Created
June 4, 2023 17:49
-
-
Save ggerganov/f1cf07a803337d638536190d124006ea to your computer and use it in GitHub Desktop.
LLaMA 7B ggml computation graph
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
20:47:34 ▶ metal ▶ 12⎘ ▶ $ ▶ ./bin/main -m ../models/7B/ggml-model-q4_0.bin --export | |
main: build = 652 (b252acb) | |
main: seed = 1685900854 | |
llama.cpp: loading model from ../models/7B/ggml-model-q4_0.bin | |
llama_model_load_internal: format = ggjt v3 (latest) | |
llama_model_load_internal: n_vocab = 32000 | |
llama_model_load_internal: n_ctx = 512 | |
llama_model_load_internal: n_embd = 4096 | |
llama_model_load_internal: n_mult = 256 | |
llama_model_load_internal: n_head = 32 | |
llama_model_load_internal: n_layer = 32 | |
llama_model_load_internal: n_rot = 128 | |
llama_model_load_internal: ftype = 2 (mostly Q4_0) | |
llama_model_load_internal: n_ff = 11008 | |
llama_model_load_internal: n_parts = 1 | |
llama_model_load_internal: model size = 7B | |
llama_model_load_internal: ggml ctx size = 0.07 MB | |
llama_model_load_internal: mem required = 5407.71 MB (+ 1026.00 MB per state) | |
. | |
llama_init_from_file: kv self size = 256.00 MB | |
system_info: n_threads = 8 / 10 | AVX = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 0 | NEON = 1 | ARM_FMA = 1 | F16C = 0 | FP16_VA = 1 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 0 | VSX = 0 | | |
magic 67676d6c | |
version 1 | |
leafs 582 | |
nodes 1188 | |
eval 831042560 | |
TYPE OP NDIMS NE0 NE1 NE2 NE3 NB0 NB1 NB2 NB3 DATA NAME | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x28960db00 layers.0.attention.wk.weight | |
q4_0 NONE 2 4096 32000 1 1 18 2304 73728000 73728000 0x280069a20 tok_embeddings.weight | |
i32 NONE 1 1 1 1 1 4 4 4 4 0x362014100 embd | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x28b10dbc0 layers.0.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362018c20 leaf_4 | |
f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0x125600100 cache_k | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362019130 leaf_6 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x289f0db40 layers.0.attention.wv.weight | |
f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0x12d600200 cache_v | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362019340 leaf_9 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2902a5d40 layers.1.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x28c941c40 layers.0.feed_forward.w2.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x28b111c00 layers.0.feed_forward.w1.weight | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x28a80db80 layers.0.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36201a280 leaf_14 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362019850 leaf_15 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x288d0dac0 layers.0.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362018810 leaf_17 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362019c60 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362019f70 leaf_19 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_20 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x28f9a1cc0 layers.0.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x28e171c80 layers.0.feed_forward.w3.weight | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x291da5e00 layers.1.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36201baa0 leaf_24 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36201bfb0 leaf_25 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x290ba5d80 layers.1.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36201c1c0 leaf_27 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x296f3df80 layers.2.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2935d9e80 layers.1.feed_forward.w2.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x291da9e40 layers.1.feed_forward.w1.weight | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2914a5dc0 layers.1.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36201d100 leaf_32 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36201c6d0 leaf_33 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x28f9a5d00 layers.1.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36201b690 leaf_35 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36201cae0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36201cdf0 leaf_37 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_38 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x296639f00 layers.1.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x294e09ec0 layers.1.feed_forward.w3.weight | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x298a3e040 layers.2.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36201e920 leaf_42 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36201ee30 leaf_43 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x29783dfc0 layers.2.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36201f040 leaf_45 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x29dbd61c0 layers.3.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x29a2720c0 layers.2.feed_forward.w2.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x298a42080 layers.2.feed_forward.w1.weight | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x29813e000 layers.2.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36201ff80 leaf_50 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36201f550 leaf_51 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x29663df40 layers.2.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36201e510 leaf_53 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36201f960 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36201fc70 leaf_55 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_56 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x29d2d2140 layers.2.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x29baa2100 layers.2.feed_forward.w3.weight | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x29f6d6280 layers.3.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x3620217a0 leaf_60 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362021cb0 leaf_61 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x29e4d6200 layers.3.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362021ec0 leaf_63 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2a486e400 layers.4.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2a0f0a300 layers.3.feed_forward.w2.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x29f6da2c0 layers.3.feed_forward.w1.weight | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x29edd6240 layers.3.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362022e00 leaf_68 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620223d0 leaf_69 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x29d2d6180 layers.3.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362021390 leaf_71 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x3620227e0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362022af0 leaf_73 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_74 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2a3f6a380 layers.3.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2a273a340 layers.3.feed_forward.w3.weight | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2a636e4c0 layers.4.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362024620 leaf_78 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362024b30 leaf_79 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2a516e440 layers.4.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362024d40 leaf_81 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2ab506640 layers.5.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2a7ba2540 layers.4.feed_forward.w2.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2a6372500 layers.4.feed_forward.w1.weight | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2a5a6e480 layers.4.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362025c80 leaf_86 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362025250 leaf_87 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2a3f6e3c0 layers.4.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362024210 leaf_89 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362025660 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362025970 leaf_91 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_92 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2aac025c0 layers.4.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2a93d2580 layers.4.feed_forward.w3.weight | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2ad006700 layers.5.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x3620274a0 leaf_96 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620279b0 leaf_97 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2abe06680 layers.5.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362027bc0 leaf_99 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2b219e880 layers.6.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2ae83a780 layers.5.feed_forward.w2.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2ad00a740 layers.5.feed_forward.w1.weight | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2ac7066c0 layers.5.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362028b00 leaf_104 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620280d0 leaf_105 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2aac06600 layers.5.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362027090 leaf_107 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x3620284e0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620287f0 leaf_109 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_110 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2b189a800 layers.5.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2b006a7c0 layers.5.feed_forward.w3.weight | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2b3c9e940 layers.6.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36202a320 leaf_114 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36202a830 leaf_115 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2b2a9e8c0 layers.6.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36202aa40 leaf_117 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2b8e36ac0 layers.7.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2b54d29c0 layers.6.feed_forward.w2.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2b3ca2980 layers.6.feed_forward.w1.weight | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2b339e900 layers.6.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36202b980 leaf_122 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36202af50 leaf_123 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2b189e840 layers.6.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362029f10 leaf_125 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36202b360 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36202b670 leaf_127 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_128 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2b8532a40 layers.6.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2b6d02a00 layers.6.feed_forward.w3.weight | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2ba936b80 layers.7.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36202d1a0 leaf_132 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36202d6b0 leaf_133 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2b9736b00 layers.7.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36202d8c0 leaf_135 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2bfaced00 layers.8.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2bc16ac00 layers.7.feed_forward.w2.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2ba93abc0 layers.7.feed_forward.w1.weight | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2ba036b40 layers.7.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36202e800 leaf_140 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36202ddd0 leaf_141 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2b8536a80 layers.7.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36202cd90 leaf_143 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36202e1e0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36202e4f0 leaf_145 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_146 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2bf1cac80 layers.7.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2bd99ac40 layers.7.feed_forward.w3.weight | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2c15cedc0 layers.8.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362030020 leaf_150 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362030530 leaf_151 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2c03ced40 layers.8.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362030740 leaf_153 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2c6766f40 layers.9.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2c2e02e40 layers.8.feed_forward.w2.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2c15d2e00 layers.8.feed_forward.w1.weight | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2c0cced80 layers.8.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362031680 leaf_158 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362030c50 leaf_159 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2bf1cecc0 layers.8.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36202fc10 leaf_161 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362031060 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362031370 leaf_163 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_164 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2c5e62ec0 layers.8.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2c4632e80 layers.8.feed_forward.w3.weight | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2c8267000 layers.9.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362032ea0 leaf_168 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620333b0 leaf_169 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2c7066f80 layers.9.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620335c0 leaf_171 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2cd3ff180 layers.10.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2c9a9b080 layers.9.feed_forward.w2.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2c826b040 layers.9.feed_forward.w1.weight | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2c7966fc0 layers.9.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362034500 leaf_176 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362033ad0 leaf_177 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2c5e66f00 layers.9.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362032a90 leaf_179 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362033ee0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620341f0 leaf_181 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_182 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2ccafb100 layers.9.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2cb2cb0c0 layers.9.feed_forward.w3.weight | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2ceeff240 layers.10.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362035d20 leaf_186 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362036230 leaf_187 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2cdcff1c0 layers.10.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362036440 leaf_189 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2d40973c0 layers.11.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2d07332c0 layers.10.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2cef03280 layers.10.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2ce5ff200 layers.10.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362037380 leaf_194 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362036950 leaf_195 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2ccaff140 layers.10.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362035910 leaf_197 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362036d60 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362037070 leaf_199 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_200 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2d3793340 layers.10.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2d1f63300 layers.10.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2d5b97480 layers.11.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362038ba0 leaf_204 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620390b0 leaf_205 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2d4997400 layers.11.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620392c0 leaf_207 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2dad2f600 layers.12.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2d73cb500 layers.11.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2d5b9b4c0 layers.11.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2d5297440 layers.11.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36203a200 leaf_212 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620397d0 leaf_213 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2d3797380 layers.11.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362038790 leaf_215 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362039be0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362039ef0 leaf_217 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_218 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2da42b580 layers.11.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2d8bfb540 layers.11.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2dc82f6c0 layers.12.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36203ba20 leaf_222 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36203bf30 leaf_223 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2db62f640 layers.12.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36203c140 leaf_225 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2e19c7840 layers.13.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2de063740 layers.12.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2dc833700 layers.12.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2dbf2f680 layers.12.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36203d080 leaf_230 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36203c650 leaf_231 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2da42f5c0 layers.12.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36203b610 leaf_233 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36203ca60 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36203cd70 leaf_235 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_236 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2e10c37c0 layers.12.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2df893780 layers.12.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2e34c7900 layers.13.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36203e8a0 leaf_240 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36203edb0 leaf_241 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2e22c7880 layers.13.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36203efc0 leaf_243 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2e865fa80 layers.14.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2e4cfb980 layers.13.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2e34cb940 layers.13.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2e2bc78c0 layers.13.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36203ff00 leaf_248 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36203f4d0 leaf_249 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2e10c7800 layers.13.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36203e490 leaf_251 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36203f8e0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36203fbf0 leaf_253 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_254 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2e7d5ba00 layers.13.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2e652b9c0 layers.13.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2ea15fb40 layers.14.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362041720 leaf_258 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362041c30 leaf_259 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2e8f5fac0 layers.14.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362041e40 leaf_261 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2ef2f7cc0 layers.15.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2eb993bc0 layers.14.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2ea163b80 layers.14.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2e985fb00 layers.14.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362042d80 leaf_266 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362042350 leaf_267 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2e7d5fa40 layers.14.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362041310 leaf_269 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362042760 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362042a70 leaf_271 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_272 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2ee9f3c40 layers.14.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2ed1c3c00 layers.14.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2f0df7d80 layers.15.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x3620445a0 leaf_276 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362044ab0 leaf_277 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2efbf7d00 layers.15.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362044cc0 leaf_279 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2f5f8ff00 layers.16.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2f262be00 layers.15.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2f0dfbdc0 layers.15.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2f04f7d40 layers.15.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362045c00 leaf_284 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620451d0 leaf_285 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2ee9f7c80 layers.15.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362044190 leaf_287 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x3620455e0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620458f0 leaf_289 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_290 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2f568be80 layers.15.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2f3e5be40 layers.15.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2f7a8ffc0 layers.16.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362047420 leaf_294 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362047930 leaf_295 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2f688ff40 layers.16.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362047b40 leaf_297 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2fcc28140 layers.17.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2f92c4040 layers.16.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2f7a94000 layers.16.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2f718ff80 layers.16.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362048a80 leaf_302 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362048050 leaf_303 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2f568fec0 layers.16.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362047010 leaf_305 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362048460 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362048770 leaf_307 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_308 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2fc3240c0 layers.16.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2faaf4080 layers.16.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2fe728200 layers.17.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36204a2a0 leaf_312 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36204a7b0 leaf_313 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2fd528180 layers.17.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36204a9c0 leaf_315 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3038c0380 layers.18.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x2fff5c280 layers.17.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x2fe72c240 layers.17.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2fde281c0 layers.17.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36204b900 leaf_320 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36204aed0 leaf_321 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x2fc328100 layers.17.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362049e90 leaf_323 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36204b2e0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36204b5f0 leaf_325 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_326 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x302fbc300 layers.17.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x30178c2c0 layers.17.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x3053c0440 layers.18.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36204d120 leaf_330 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36204d630 leaf_331 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3041c03c0 layers.18.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36204d840 leaf_333 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x30a5585c0 layers.19.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x306bf44c0 layers.18.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x3053c4480 layers.18.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x304ac0400 layers.18.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36204e780 leaf_338 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36204dd50 leaf_339 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x302fc0340 layers.18.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36204cd10 leaf_341 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36204e160 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36204e470 leaf_343 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_344 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x309c54540 layers.18.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x308424500 layers.18.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x30c058680 layers.19.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36204ffa0 leaf_348 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620504b0 leaf_349 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x30ae58600 layers.19.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620506c0 leaf_351 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3111f0800 layers.20.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x30d88c700 layers.19.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x30c05c6c0 layers.19.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x30b758640 layers.19.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362051600 leaf_356 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362050bd0 leaf_357 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x309c58580 layers.19.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36204fb90 leaf_359 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362050fe0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620512f0 leaf_361 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_362 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x3108ec780 layers.19.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x30f0bc740 layers.19.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x312cf08c0 layers.20.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362052e20 leaf_366 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362053330 leaf_367 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x311af0840 layers.20.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362053540 leaf_369 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x317e88a40 layers.21.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x314524940 layers.20.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x312cf4900 layers.20.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3123f0880 layers.20.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362054480 leaf_374 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362053a50 leaf_375 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3108f07c0 layers.20.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362052a10 leaf_377 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362053e60 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362054170 leaf_379 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_380 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x3175849c0 layers.20.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x315d54980 layers.20.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x319988b00 layers.21.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362055ca0 leaf_384 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620561b0 leaf_385 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x318788a80 layers.21.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620563c0 leaf_387 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x31eb20c80 layers.22.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x31b1bcb80 layers.21.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x31998cb40 layers.21.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x319088ac0 layers.21.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362057300 leaf_392 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620568d0 leaf_393 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x317588a00 layers.21.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362055890 leaf_395 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362056ce0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362056ff0 leaf_397 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_398 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x31e21cc00 layers.21.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x31c9ecbc0 layers.21.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x320620d40 layers.22.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362058b20 leaf_402 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362059030 leaf_403 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x31f420cc0 layers.22.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362059240 leaf_405 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3257b8ec0 layers.23.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x321e54dc0 layers.22.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x320624d80 layers.22.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x31fd20d00 layers.22.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36205a180 leaf_410 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362059750 leaf_411 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x31e220c40 layers.22.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362058710 leaf_413 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362059b60 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362059e70 leaf_415 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_416 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x324eb4e40 layers.22.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x323684e00 layers.22.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x3272b8f80 layers.23.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36205b9a0 leaf_420 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36205beb0 leaf_421 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3260b8f00 layers.23.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36205c0c0 leaf_423 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x32c451100 layers.24.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x328aed000 layers.23.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x3272bcfc0 layers.23.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3269b8f40 layers.23.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36205d000 leaf_428 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36205c5d0 leaf_429 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x324eb8e80 layers.23.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36205b590 leaf_431 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36205c9e0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36205ccf0 leaf_433 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_434 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x32bb4d080 layers.23.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x32a31d040 layers.23.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x32df511c0 layers.24.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36205e820 leaf_438 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36205ed30 leaf_439 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x32cd51140 layers.24.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36205ef40 leaf_441 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3330e9340 layers.25.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x32f785240 layers.24.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x32df55200 layers.24.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x32d651180 layers.24.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36205fe80 leaf_446 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36205f450 leaf_447 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x32bb510c0 layers.24.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36205e410 leaf_449 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36205f860 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36205fb70 leaf_451 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_452 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x3327e52c0 layers.24.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x330fb5280 layers.24.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x334be9400 layers.25.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x3620616a0 leaf_456 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362061bb0 leaf_457 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3339e9380 layers.25.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362061dc0 leaf_459 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x339d81580 layers.26.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x33641d480 layers.25.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x334bed440 layers.25.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3342e93c0 layers.25.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362062d00 leaf_464 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620622d0 leaf_465 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3327e9300 layers.25.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362061290 leaf_467 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x3620626e0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620629f0 leaf_469 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_470 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x33947d500 layers.25.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x337c4d4c0 layers.25.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x33b881640 layers.26.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362064520 leaf_474 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362064a30 leaf_475 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x33a6815c0 layers.26.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362064c40 leaf_477 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x340a197c0 layers.27.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x33d0b56c0 layers.26.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x33b885680 layers.26.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x33af81600 layers.26.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362065b80 leaf_482 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362065150 leaf_483 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x339481540 layers.26.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362064110 leaf_485 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362065560 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362065870 leaf_487 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_488 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x340115740 layers.26.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x33e8e5700 layers.26.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x342519880 layers.27.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x3620673a0 leaf_492 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620678b0 leaf_493 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x341319800 layers.27.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362067ac0 leaf_495 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3476b1a00 layers.28.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x343d4d900 layers.27.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x34251d8c0 layers.27.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x341c19840 layers.27.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362068a00 leaf_500 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362067fd0 leaf_501 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x340119780 layers.27.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362066f90 leaf_503 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x3620683e0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620686f0 leaf_505 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_506 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x346dad980 layers.27.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x34557d940 layers.27.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x3491b1ac0 layers.28.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36206a220 leaf_510 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36206a730 leaf_511 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x347fb1a40 layers.28.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36206a940 leaf_513 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x34e349c40 layers.29.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x34a9e5b40 layers.28.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x3491b5b00 layers.28.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3488b1a80 layers.28.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36206b880 leaf_518 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36206ae50 leaf_519 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x346db19c0 layers.28.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362069e10 leaf_521 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36206b260 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36206b570 leaf_523 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_524 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x34da45bc0 layers.28.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x34c215b80 layers.28.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x34fe49d00 layers.29.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36206d0a0 leaf_528 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36206d5b0 leaf_529 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x34ec49c80 layers.29.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36206d7c0 leaf_531 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x354fe1e80 layers.30.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x35167dd80 layers.29.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x34fe4dd40 layers.29.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x34f549cc0 layers.29.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36206e700 leaf_536 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36206dcd0 leaf_537 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x34da49c00 layers.29.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36206cc90 leaf_539 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x36206e0e0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x36206e3f0 leaf_541 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_542 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x3546dde00 layers.29.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x352eaddc0 layers.29.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x356ae1f40 layers.30.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36206ff20 leaf_546 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362070430 leaf_547 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3558e1ec0 layers.30.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362070640 leaf_549 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x35bc7a0c0 layers.31.attention.wk.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x358315fc0 layers.30.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x356ae5f80 layers.30.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3561e1f00 layers.30.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362071580 leaf_554 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362070b50 leaf_555 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x3546e1e40 layers.30.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x36206fb10 leaf_557 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362070f60 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362071270 leaf_559 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_560 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x35b376040 layers.30.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x359b46000 layers.30.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x35d77a180 layers.31.attention_norm.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362072da0 leaf_564 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620732b0 leaf_565 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x35c57a100 layers.31.attention.wv.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620734c0 leaf_567 | |
q4_0 NONE 2 4096 32000 1 1 18 2304 73728000 73728000 0x2846bda80 output.weight | |
q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0x35efae200 layers.31.feed_forward.w2.weigh | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x35d77e1c0 layers.31.feed_forward.w1.weigh | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x35ce7a140 layers.31.attention.wo.weight | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x362074400 leaf_572 | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620739d0 leaf_573 | |
q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0x35b37a080 layers.31.attention.wq.weight | |
i32 NONE 1 3 1 1 1 4 12 12 12 0x362072990 leaf_575 | |
f32 NONE 1 1 1 1 1 4 4 4 4 0x362073de0 1/sqrt(n_embd/n_head) | |
i32 NONE 1 2 1 1 1 4 8 8 8 0x3620740f0 leaf_577 | |
f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0x1051cc000 leaf_578 | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x36200e280 layers.31.ffn_norm.weight | |
q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0x3607de240 layers.31.feed_forward.w3.weigh | |
f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0x2846b9a40 norm.weight | |
ARG TYPE OP NDIMS NE0 NE1 NE2 NE3 NB0 NB1 NB2 NB3 NTASKS DATA NAME | |
DST f32 GET_ROWS 2 4096 1 1 1 4 16384 16384 16384 1 0x362014210 node_0 | |
SRC0 q4_0 NONE 2 4096 32000 1 1 18 2304 73728000 73728000 0 0x280069a20 tok_embeddings.weight | |
SRC1 i32 NONE 1 1 1 1 1 4 4 4 4 0 0x362014100 embd | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1 | |
SRC0 f32 GET_ROWS 2 4096 1 1 1 4 16384 16384 16384 1 0x362014210 node_0 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_2 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x28b10dbc0 layers.0.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_3 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x28960db00 layers.0.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_2 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_4 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_3 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_4 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362018c20 leaf_4 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1259fe100 node_6 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362019130 leaf_6 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1259fe100 node_7 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1259fe100 node_6 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_8 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x289f0db40 layers.0.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_2 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_9 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_8 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_9 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12d6005fe node_11 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362019340 leaf_9 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12d6005fe node_12 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12d6005fe node_11 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12d600200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36201a280 leaf_14 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x125600100 node_14 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362019850 leaf_15 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x125600100 node_15 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x125600100 node_14 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x125600100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x125600100 node_15 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_17 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x288d0dac0 layers.0.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_2 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_18 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_17 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_18 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362018810 leaf_17 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x125600100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362019c60 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362019f70 leaf_19 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12d600200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_20 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_28 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x28a80db80 layers.0.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_29 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_28 | |
SRC1 f32 GET_ROWS 2 4096 1 1 1 4 16384 16384 16384 1 0x362014210 node_0 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_30 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_29 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_31 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_30 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x28f9a1cc0 layers.0.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_32 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x28b111c00 layers.0.feed_forward.w1.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_31 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_33 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_32 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_34 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x28e171c80 layers.0.feed_forward.w3.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_31 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_35 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_33 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_34 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_36 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x28c941c40 layers.0.feed_forward.w2.weight | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_35 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_37 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_36 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_29 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_38 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_37 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_39 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_38 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x291da5e00 layers.1.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_40 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2902a5d40 layers.1.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_39 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_41 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_40 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_41 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36201baa0 leaf_24 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x125dfe100 node_43 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36201bfb0 leaf_25 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x125dfe100 node_44 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x125dfe100 node_43 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_45 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x290ba5d80 layers.1.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_39 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_46 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_45 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_46 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12da005fe node_48 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36201c1c0 leaf_27 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12da005fe node_49 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12da005fe node_48 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12da00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36201d100 leaf_32 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x125a00100 node_51 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36201c6d0 leaf_33 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x125a00100 node_52 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x125a00100 node_51 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x125a00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x125a00100 node_52 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_54 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x28f9a5d00 layers.1.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_39 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_55 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_54 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_55 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36201b690 leaf_35 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x125a00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36201cae0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36201cdf0 leaf_37 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12da00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_38 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_65 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2914a5dc0 layers.1.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_66 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_65 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_37 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_67 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_66 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_68 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_67 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x296639f00 layers.1.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_69 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x291da9e40 layers.1.feed_forward.w1.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_68 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_70 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_69 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_71 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x294e09ec0 layers.1.feed_forward.w3.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_68 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_72 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_70 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_71 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_73 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2935d9e80 layers.1.feed_forward.w2.weight | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_72 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_74 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_73 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_66 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_75 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_74 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_76 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_75 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x298a3e040 layers.2.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_77 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x296f3df80 layers.2.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_76 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_78 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_77 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_78 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36201e920 leaf_42 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1261fe100 node_80 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36201ee30 leaf_43 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1261fe100 node_81 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1261fe100 node_80 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_82 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x29783dfc0 layers.2.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_76 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_83 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_82 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_83 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12de005fe node_85 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36201f040 leaf_45 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12de005fe node_86 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12de005fe node_85 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12de00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36201ff80 leaf_50 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x125e00100 node_88 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36201f550 leaf_51 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x125e00100 node_89 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x125e00100 node_88 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x125e00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x125e00100 node_89 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_91 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x29663df40 layers.2.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_76 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_92 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_91 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_92 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36201e510 leaf_53 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x125e00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36201f960 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36201fc70 leaf_55 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12de00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_56 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_102 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x29813e000 layers.2.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_103 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_102 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_74 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_104 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_103 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_105 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_104 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x29d2d2140 layers.2.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_106 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x298a42080 layers.2.feed_forward.w1.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_105 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_107 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_106 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_108 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x29baa2100 layers.2.feed_forward.w3.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_105 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_109 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_107 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_108 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_110 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x29a2720c0 layers.2.feed_forward.w2.weight | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_109 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_111 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_110 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_103 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_112 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_111 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_113 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_112 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x29f6d6280 layers.3.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_114 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x29dbd61c0 layers.3.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_113 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_115 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_114 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_115 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x3620217a0 leaf_60 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1265fe100 node_117 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362021cb0 leaf_61 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1265fe100 node_118 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1265fe100 node_117 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_119 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x29e4d6200 layers.3.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_113 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_120 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_119 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_120 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12e2005fe node_122 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362021ec0 leaf_63 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12e2005fe node_123 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12e2005fe node_122 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12e200200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362022e00 leaf_68 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x126200100 node_125 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620223d0 leaf_69 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x126200100 node_126 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x126200100 node_125 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x126200100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x126200100 node_126 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_128 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x29d2d6180 layers.3.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_113 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_129 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_128 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_129 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362021390 leaf_71 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x126200100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x3620227e0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362022af0 leaf_73 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12e200200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_74 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_139 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x29edd6240 layers.3.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_140 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_139 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_111 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_141 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_140 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_142 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_141 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2a3f6a380 layers.3.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_143 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x29f6da2c0 layers.3.feed_forward.w1.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_142 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_144 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_143 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_145 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2a273a340 layers.3.feed_forward.w3.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_142 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_146 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_144 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_145 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_147 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2a0f0a300 layers.3.feed_forward.w2.weight | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_146 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_148 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_147 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_140 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_149 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_148 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_150 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_149 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2a636e4c0 layers.4.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_151 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2a486e400 layers.4.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_150 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_152 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_151 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_152 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362024620 leaf_78 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1269fe100 node_154 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362024b30 leaf_79 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1269fe100 node_155 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1269fe100 node_154 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_156 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2a516e440 layers.4.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_150 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_157 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_156 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_157 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12e6005fe node_159 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362024d40 leaf_81 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12e6005fe node_160 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12e6005fe node_159 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12e600200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362025c80 leaf_86 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x126600100 node_162 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362025250 leaf_87 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x126600100 node_163 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x126600100 node_162 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x126600100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x126600100 node_163 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_165 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2a3f6e3c0 layers.4.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_150 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_166 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_165 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_166 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362024210 leaf_89 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x126600100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362025660 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362025970 leaf_91 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12e600200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_92 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_176 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2a5a6e480 layers.4.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_177 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_176 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_148 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_178 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_177 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_179 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_178 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2aac025c0 layers.4.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_180 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2a6372500 layers.4.feed_forward.w1.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_179 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_181 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_180 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_182 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2a93d2580 layers.4.feed_forward.w3.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_179 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_183 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_181 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_182 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_184 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2a7ba2540 layers.4.feed_forward.w2.weight | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_183 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_185 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_184 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_177 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_186 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_185 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_187 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_186 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2ad006700 layers.5.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_188 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2ab506640 layers.5.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_187 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_189 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_188 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_189 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x3620274a0 leaf_96 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x126dfe100 node_191 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620279b0 leaf_97 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x126dfe100 node_192 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x126dfe100 node_191 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_193 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2abe06680 layers.5.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_187 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_194 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_193 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_194 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12ea005fe node_196 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362027bc0 leaf_99 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12ea005fe node_197 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12ea005fe node_196 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12ea00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362028b00 leaf_104 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x126a00100 node_199 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620280d0 leaf_105 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x126a00100 node_200 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x126a00100 node_199 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x126a00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x126a00100 node_200 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_202 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2aac06600 layers.5.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_187 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_203 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_202 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_203 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362027090 leaf_107 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x126a00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x3620284e0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620287f0 leaf_109 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12ea00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_110 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_213 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2ac7066c0 layers.5.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_214 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_213 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_185 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_215 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_214 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_216 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_215 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2b189a800 layers.5.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_217 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2ad00a740 layers.5.feed_forward.w1.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_216 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_218 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_217 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_219 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2b006a7c0 layers.5.feed_forward.w3.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_216 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_220 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_218 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_219 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_221 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2ae83a780 layers.5.feed_forward.w2.weight | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_220 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_222 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_221 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_214 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_223 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_222 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_224 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_223 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2b3c9e940 layers.6.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_225 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2b219e880 layers.6.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_224 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_226 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_225 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_226 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36202a320 leaf_114 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1271fe100 node_228 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36202a830 leaf_115 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1271fe100 node_229 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1271fe100 node_228 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_230 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2b2a9e8c0 layers.6.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_224 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_231 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_230 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_231 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12ee005fe node_233 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36202aa40 leaf_117 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12ee005fe node_234 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12ee005fe node_233 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12ee00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36202b980 leaf_122 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x126e00100 node_236 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36202af50 leaf_123 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x126e00100 node_237 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x126e00100 node_236 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x126e00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x126e00100 node_237 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_239 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2b189e840 layers.6.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_224 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_240 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_239 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_240 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362029f10 leaf_125 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x126e00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36202b360 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36202b670 leaf_127 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12ee00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_128 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_250 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2b339e900 layers.6.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_251 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_250 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_222 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_252 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_251 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_253 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_252 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2b8532a40 layers.6.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_254 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2b3ca2980 layers.6.feed_forward.w1.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_253 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_255 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_254 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_256 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2b6d02a00 layers.6.feed_forward.w3.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_253 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_257 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_255 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_256 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_258 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2b54d29c0 layers.6.feed_forward.w2.weight | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_257 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_259 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_258 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_251 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_260 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_259 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_261 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_260 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2ba936b80 layers.7.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_262 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2b8e36ac0 layers.7.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_261 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_263 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_262 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_263 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36202d1a0 leaf_132 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1275fe100 node_265 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36202d6b0 leaf_133 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1275fe100 node_266 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1275fe100 node_265 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_267 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2b9736b00 layers.7.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_261 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_268 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_267 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_268 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12f2005fe node_270 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36202d8c0 leaf_135 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12f2005fe node_271 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12f2005fe node_270 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12f200200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36202e800 leaf_140 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x127200100 node_273 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36202ddd0 leaf_141 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x127200100 node_274 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x127200100 node_273 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x127200100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x127200100 node_274 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_276 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2b8536a80 layers.7.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_261 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_277 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_276 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_277 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36202cd90 leaf_143 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x127200100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36202e1e0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36202e4f0 leaf_145 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12f200200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_146 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_287 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2ba036b40 layers.7.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_288 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_287 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_259 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_289 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_288 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_290 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_289 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2bf1cac80 layers.7.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_291 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2ba93abc0 layers.7.feed_forward.w1.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_290 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_292 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_291 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_293 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2bd99ac40 layers.7.feed_forward.w3.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_290 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_294 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_292 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_293 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_295 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2bc16ac00 layers.7.feed_forward.w2.weight | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_294 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_296 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_295 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_288 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_297 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_296 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_298 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_297 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2c15cedc0 layers.8.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_299 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2bfaced00 layers.8.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_298 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_300 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_299 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_300 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362030020 leaf_150 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1279fe100 node_302 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362030530 leaf_151 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1279fe100 node_303 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1279fe100 node_302 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_304 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2c03ced40 layers.8.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_298 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_305 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_304 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_305 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12f6005fe node_307 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362030740 leaf_153 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12f6005fe node_308 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12f6005fe node_307 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12f600200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362031680 leaf_158 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x127600100 node_310 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362030c50 leaf_159 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x127600100 node_311 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x127600100 node_310 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x127600100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x127600100 node_311 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_313 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2bf1cecc0 layers.8.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_298 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_314 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_313 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_314 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36202fc10 leaf_161 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x127600100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362031060 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362031370 leaf_163 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12f600200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_164 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_324 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2c0cced80 layers.8.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_325 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_324 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_296 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_326 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_325 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_327 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_326 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2c5e62ec0 layers.8.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_328 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2c15d2e00 layers.8.feed_forward.w1.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_327 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_329 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_328 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_330 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2c4632e80 layers.8.feed_forward.w3.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_327 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_331 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_329 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_330 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_332 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2c2e02e40 layers.8.feed_forward.w2.weight | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_331 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_333 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_332 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_325 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_334 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_333 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_335 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_334 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2c8267000 layers.9.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_336 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2c6766f40 layers.9.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_335 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_337 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_336 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_337 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362032ea0 leaf_168 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x127dfe100 node_339 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620333b0 leaf_169 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x127dfe100 node_340 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x127dfe100 node_339 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_341 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2c7066f80 layers.9.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_335 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_342 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_341 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_342 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12fa005fe node_344 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620335c0 leaf_171 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12fa005fe node_345 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12fa005fe node_344 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12fa00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362034500 leaf_176 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x127a00100 node_347 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362033ad0 leaf_177 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x127a00100 node_348 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x127a00100 node_347 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x127a00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x127a00100 node_348 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_350 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2c5e66f00 layers.9.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_335 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_351 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_350 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_351 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362032a90 leaf_179 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x127a00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362033ee0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620341f0 leaf_181 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12fa00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_182 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_361 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2c7966fc0 layers.9.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_362 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_361 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_333 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_363 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_362 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_364 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_363 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2ccafb100 layers.9.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_365 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2c826b040 layers.9.feed_forward.w1.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_364 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_366 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_365 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_367 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2cb2cb0c0 layers.9.feed_forward.w3.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_364 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_368 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_366 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_367 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_369 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2c9a9b080 layers.9.feed_forward.w2.weight | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_368 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_370 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_369 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_362 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_371 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_370 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_372 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_371 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2ceeff240 layers.10.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_373 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2cd3ff180 layers.10.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_372 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_374 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_373 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_374 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362035d20 leaf_186 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1281fe100 node_376 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362036230 leaf_187 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1281fe100 node_377 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1281fe100 node_376 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_378 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2cdcff1c0 layers.10.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_372 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_379 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_378 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_379 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12fe005fe node_381 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362036440 leaf_189 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12fe005fe node_382 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x12fe005fe node_381 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12fe00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362037380 leaf_194 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x127e00100 node_384 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362036950 leaf_195 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x127e00100 node_385 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x127e00100 node_384 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x127e00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x127e00100 node_385 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_387 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2ccaff140 layers.10.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_372 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_388 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_387 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_388 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362035910 leaf_197 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x127e00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362036d60 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362037070 leaf_199 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x12fe00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_200 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_398 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2ce5ff200 layers.10.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_399 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_398 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_370 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_400 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_399 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_401 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_400 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2d3793340 layers.10.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_402 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2cef03280 layers.10.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_401 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_403 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_402 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_404 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2d1f63300 layers.10.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_401 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_405 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_403 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_404 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_406 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2d07332c0 layers.10.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_405 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_407 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_406 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_399 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_408 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_407 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_409 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_408 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2d5b97480 layers.11.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_410 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2d40973c0 layers.11.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_409 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_411 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_410 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_411 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362038ba0 leaf_204 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1285fe100 node_413 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620390b0 leaf_205 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1285fe100 node_414 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1285fe100 node_413 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_415 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2d4997400 layers.11.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_409 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_416 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_415 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_416 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1302005fe node_418 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620392c0 leaf_207 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1302005fe node_419 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1302005fe node_418 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x130200200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36203a200 leaf_212 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x128200100 node_421 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620397d0 leaf_213 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x128200100 node_422 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x128200100 node_421 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x128200100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x128200100 node_422 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_424 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2d3797380 layers.11.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_409 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_425 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_424 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_425 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362038790 leaf_215 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x128200100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362039be0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362039ef0 leaf_217 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x130200200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_218 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_435 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2d5297440 layers.11.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_436 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_435 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_407 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_437 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_436 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_438 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_437 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2da42b580 layers.11.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_439 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2d5b9b4c0 layers.11.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_438 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_440 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_439 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_441 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2d8bfb540 layers.11.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_438 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_442 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_440 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_441 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_443 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2d73cb500 layers.11.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_442 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_444 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_443 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_436 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_445 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_444 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_446 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_445 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2dc82f6c0 layers.12.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_447 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2dad2f600 layers.12.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_446 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_448 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_447 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_448 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36203ba20 leaf_222 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1289fe100 node_450 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36203bf30 leaf_223 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1289fe100 node_451 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1289fe100 node_450 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_452 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2db62f640 layers.12.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_446 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_453 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_452 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_453 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1306005fe node_455 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36203c140 leaf_225 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1306005fe node_456 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1306005fe node_455 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x130600200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36203d080 leaf_230 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x128600100 node_458 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36203c650 leaf_231 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x128600100 node_459 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x128600100 node_458 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x128600100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x128600100 node_459 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_461 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2da42f5c0 layers.12.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_446 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_462 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_461 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_462 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36203b610 leaf_233 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x128600100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36203ca60 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36203cd70 leaf_235 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x130600200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_236 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_472 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2dbf2f680 layers.12.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_473 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_472 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_444 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_474 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_473 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_475 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_474 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2e10c37c0 layers.12.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_476 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2dc833700 layers.12.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_475 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_477 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_476 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_478 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2df893780 layers.12.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_475 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_479 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_477 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_478 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_480 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2de063740 layers.12.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_479 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_481 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_480 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_473 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_482 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_481 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_483 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_482 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2e34c7900 layers.13.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_484 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2e19c7840 layers.13.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_483 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_485 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_484 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_485 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36203e8a0 leaf_240 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x128dfe100 node_487 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36203edb0 leaf_241 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x128dfe100 node_488 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x128dfe100 node_487 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_489 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2e22c7880 layers.13.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_483 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_490 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_489 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_490 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x130a005fe node_492 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36203efc0 leaf_243 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x130a005fe node_493 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x130a005fe node_492 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x130a00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36203ff00 leaf_248 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x128a00100 node_495 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36203f4d0 leaf_249 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x128a00100 node_496 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x128a00100 node_495 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x128a00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x128a00100 node_496 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_498 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2e10c7800 layers.13.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_483 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_499 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_498 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_499 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36203e490 leaf_251 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x128a00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36203f8e0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36203fbf0 leaf_253 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x130a00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_254 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_509 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2e2bc78c0 layers.13.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_510 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_509 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_481 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_511 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_510 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_512 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_511 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2e7d5ba00 layers.13.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_513 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2e34cb940 layers.13.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_512 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_514 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_513 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_515 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2e652b9c0 layers.13.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_512 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_516 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_514 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_515 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_517 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2e4cfb980 layers.13.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_516 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_518 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_517 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_510 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_519 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_518 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_520 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_519 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2ea15fb40 layers.14.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_521 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2e865fa80 layers.14.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_520 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_522 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_521 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_522 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362041720 leaf_258 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1291fe100 node_524 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362041c30 leaf_259 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1291fe100 node_525 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1291fe100 node_524 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_526 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2e8f5fac0 layers.14.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_520 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_527 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_526 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_527 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x130e005fe node_529 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362041e40 leaf_261 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x130e005fe node_530 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x130e005fe node_529 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x130e00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362042d80 leaf_266 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x128e00100 node_532 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362042350 leaf_267 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x128e00100 node_533 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x128e00100 node_532 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x128e00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x128e00100 node_533 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_535 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2e7d5fa40 layers.14.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_520 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_536 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_535 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_536 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362041310 leaf_269 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x128e00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362042760 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362042a70 leaf_271 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x130e00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_272 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_546 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2e985fb00 layers.14.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_547 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_546 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_518 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_548 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_547 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_549 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_548 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2ee9f3c40 layers.14.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_550 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2ea163b80 layers.14.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_549 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_551 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_550 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_552 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2ed1c3c00 layers.14.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_549 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_553 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_551 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_552 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_554 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2eb993bc0 layers.14.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_553 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_555 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_554 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_547 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_556 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_555 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_557 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_556 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2f0df7d80 layers.15.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_558 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2ef2f7cc0 layers.15.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_557 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_559 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_558 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_559 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x3620445a0 leaf_276 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1295fe100 node_561 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362044ab0 leaf_277 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1295fe100 node_562 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1295fe100 node_561 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_563 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2efbf7d00 layers.15.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_557 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_564 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_563 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_564 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1312005fe node_566 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362044cc0 leaf_279 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1312005fe node_567 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1312005fe node_566 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x131200200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362045c00 leaf_284 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x129200100 node_569 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620451d0 leaf_285 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x129200100 node_570 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x129200100 node_569 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x129200100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x129200100 node_570 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_572 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2ee9f7c80 layers.15.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_557 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_573 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_572 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_573 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362044190 leaf_287 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x129200100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x3620455e0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620458f0 leaf_289 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x131200200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_290 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_583 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2f04f7d40 layers.15.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_584 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_583 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_555 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_585 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_584 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_586 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_585 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2f568be80 layers.15.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_587 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2f0dfbdc0 layers.15.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_586 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_588 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_587 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_589 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2f3e5be40 layers.15.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_586 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_590 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_588 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_589 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_591 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2f262be00 layers.15.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_590 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_592 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_591 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_584 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_593 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_592 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_594 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_593 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2f7a8ffc0 layers.16.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_595 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2f5f8ff00 layers.16.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_594 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_596 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_595 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_596 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362047420 leaf_294 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1299fe100 node_598 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362047930 leaf_295 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x1299fe100 node_599 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x1299fe100 node_598 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_600 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2f688ff40 layers.16.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_594 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_601 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_600 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_601 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1316005fe node_603 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362047b40 leaf_297 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1316005fe node_604 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1316005fe node_603 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x131600200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362048a80 leaf_302 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x129600100 node_606 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362048050 leaf_303 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x129600100 node_607 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x129600100 node_606 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x129600100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x129600100 node_607 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_609 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2f568fec0 layers.16.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_594 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_610 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_609 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_610 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362047010 leaf_305 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x129600100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362048460 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362048770 leaf_307 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x131600200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_308 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_620 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2f718ff80 layers.16.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_621 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_620 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_592 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_622 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_621 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_623 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_622 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2fc3240c0 layers.16.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_624 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2f7a94000 layers.16.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_623 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_625 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_624 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_626 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2faaf4080 layers.16.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_623 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_627 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_625 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_626 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_628 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2f92c4040 layers.16.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_627 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_629 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_628 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_621 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_630 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_629 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_631 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_630 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2fe728200 layers.17.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_632 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2fcc28140 layers.17.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_631 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_633 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_632 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_633 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36204a2a0 leaf_312 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x129dfe100 node_635 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36204a7b0 leaf_313 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x129dfe100 node_636 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x129dfe100 node_635 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_637 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2fd528180 layers.17.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_631 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_638 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_637 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_638 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x131a005fe node_640 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36204a9c0 leaf_315 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x131a005fe node_641 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x131a005fe node_640 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x131a00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36204b900 leaf_320 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x129a00100 node_643 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36204aed0 leaf_321 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x129a00100 node_644 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x129a00100 node_643 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x129a00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x129a00100 node_644 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_646 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2fc328100 layers.17.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_631 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_647 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_646 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_647 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362049e90 leaf_323 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x129a00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36204b2e0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36204b5f0 leaf_325 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x131a00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_326 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_657 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x2fde281c0 layers.17.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_658 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_657 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_629 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_659 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_658 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_660 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_659 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x302fbc300 layers.17.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_661 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x2fe72c240 layers.17.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_660 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_662 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_661 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_663 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x30178c2c0 layers.17.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_660 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_664 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_662 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_663 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_665 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x2fff5c280 layers.17.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_664 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_666 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_665 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_658 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_667 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_666 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_668 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_667 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x3053c0440 layers.18.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_669 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3038c0380 layers.18.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_668 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_670 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_669 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_670 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36204d120 leaf_330 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12a1fe100 node_672 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36204d630 leaf_331 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12a1fe100 node_673 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12a1fe100 node_672 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_674 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3041c03c0 layers.18.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_668 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_675 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_674 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_675 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x131e005fe node_677 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36204d840 leaf_333 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x131e005fe node_678 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x131e005fe node_677 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x131e00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36204e780 leaf_338 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x129e00100 node_680 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36204dd50 leaf_339 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x129e00100 node_681 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x129e00100 node_680 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x129e00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x129e00100 node_681 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_683 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x302fc0340 layers.18.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_668 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_684 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_683 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_684 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36204cd10 leaf_341 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x129e00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36204e160 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36204e470 leaf_343 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x131e00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_344 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_694 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x304ac0400 layers.18.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_695 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_694 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_666 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_696 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_695 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_697 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_696 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x309c54540 layers.18.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_698 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x3053c4480 layers.18.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_697 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_699 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_698 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_700 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x308424500 layers.18.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_697 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_701 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_699 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_700 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_702 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x306bf44c0 layers.18.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_701 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_703 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_702 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_695 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_704 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_703 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_705 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_704 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x30c058680 layers.19.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_706 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x30a5585c0 layers.19.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_705 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_707 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_706 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_707 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36204ffa0 leaf_348 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12a5fe100 node_709 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620504b0 leaf_349 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12a5fe100 node_710 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12a5fe100 node_709 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_711 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x30ae58600 layers.19.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_705 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_712 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_711 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_712 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1322005fe node_714 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620506c0 leaf_351 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1322005fe node_715 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1322005fe node_714 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x132200200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362051600 leaf_356 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12a200100 node_717 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362050bd0 leaf_357 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12a200100 node_718 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12a200100 node_717 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12a200100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12a200100 node_718 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_720 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x309c58580 layers.19.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_705 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_721 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_720 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_721 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36204fb90 leaf_359 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12a200100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362050fe0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620512f0 leaf_361 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x132200200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_362 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_731 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x30b758640 layers.19.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_732 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_731 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_703 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_733 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_732 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_734 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_733 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x3108ec780 layers.19.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_735 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x30c05c6c0 layers.19.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_734 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_736 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_735 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_737 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x30f0bc740 layers.19.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_734 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_738 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_736 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_737 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_739 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x30d88c700 layers.19.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_738 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_740 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_739 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_732 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_741 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_740 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_742 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_741 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x312cf08c0 layers.20.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_743 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3111f0800 layers.20.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_742 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_744 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_743 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_744 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362052e20 leaf_366 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12a9fe100 node_746 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362053330 leaf_367 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12a9fe100 node_747 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12a9fe100 node_746 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_748 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x311af0840 layers.20.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_742 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_749 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_748 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_749 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1326005fe node_751 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362053540 leaf_369 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1326005fe node_752 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1326005fe node_751 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x132600200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362054480 leaf_374 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12a600100 node_754 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362053a50 leaf_375 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12a600100 node_755 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12a600100 node_754 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12a600100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12a600100 node_755 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_757 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3108f07c0 layers.20.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_742 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_758 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_757 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_758 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362052a10 leaf_377 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12a600100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362053e60 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362054170 leaf_379 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x132600200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_380 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_768 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3123f0880 layers.20.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_769 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_768 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_740 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_770 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_769 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_771 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_770 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x3175849c0 layers.20.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_772 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x312cf4900 layers.20.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_771 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_773 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_772 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_774 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x315d54980 layers.20.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_771 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_775 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_773 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_774 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_776 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x314524940 layers.20.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_775 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_777 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_776 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_769 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_778 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_777 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_779 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_778 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x319988b00 layers.21.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_780 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x317e88a40 layers.21.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_779 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_781 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_780 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_781 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362055ca0 leaf_384 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12adfe100 node_783 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620561b0 leaf_385 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12adfe100 node_784 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12adfe100 node_783 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_785 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x318788a80 layers.21.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_779 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_786 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_785 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_786 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x132a005fe node_788 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620563c0 leaf_387 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x132a005fe node_789 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x132a005fe node_788 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x132a00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362057300 leaf_392 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12aa00100 node_791 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620568d0 leaf_393 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12aa00100 node_792 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12aa00100 node_791 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12aa00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12aa00100 node_792 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_794 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x317588a00 layers.21.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_779 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_795 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_794 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_795 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362055890 leaf_395 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12aa00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362056ce0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362056ff0 leaf_397 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x132a00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_398 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_805 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x319088ac0 layers.21.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_806 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_805 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_777 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_807 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_806 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_808 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_807 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x31e21cc00 layers.21.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_809 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x31998cb40 layers.21.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_808 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_810 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_809 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_811 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x31c9ecbc0 layers.21.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_808 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_812 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_810 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_811 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_813 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x31b1bcb80 layers.21.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_812 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_814 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_813 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_806 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_815 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_814 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_816 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_815 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x320620d40 layers.22.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_817 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x31eb20c80 layers.22.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_816 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_818 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_817 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_818 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362058b20 leaf_402 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12b1fe100 node_820 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362059030 leaf_403 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12b1fe100 node_821 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12b1fe100 node_820 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_822 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x31f420cc0 layers.22.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_816 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_823 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_822 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_823 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x132e005fe node_825 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362059240 leaf_405 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x132e005fe node_826 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x132e005fe node_825 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x132e00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36205a180 leaf_410 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12ae00100 node_828 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362059750 leaf_411 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12ae00100 node_829 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12ae00100 node_828 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12ae00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12ae00100 node_829 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_831 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x31e220c40 layers.22.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_816 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_832 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_831 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_832 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362058710 leaf_413 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12ae00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362059b60 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362059e70 leaf_415 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x132e00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_416 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_842 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x31fd20d00 layers.22.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_843 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_842 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_814 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_844 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_843 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_845 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_844 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x324eb4e40 layers.22.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_846 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x320624d80 layers.22.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_845 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_847 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_846 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_848 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x323684e00 layers.22.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_845 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_849 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_847 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_848 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_850 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x321e54dc0 layers.22.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_849 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_851 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_850 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_843 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_852 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_851 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_853 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_852 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x3272b8f80 layers.23.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_854 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3257b8ec0 layers.23.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_853 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_855 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_854 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_855 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36205b9a0 leaf_420 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12b5fe100 node_857 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36205beb0 leaf_421 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12b5fe100 node_858 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12b5fe100 node_857 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_859 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3260b8f00 layers.23.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_853 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_860 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_859 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_860 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1332005fe node_862 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36205c0c0 leaf_423 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1332005fe node_863 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1332005fe node_862 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x133200200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36205d000 leaf_428 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12b200100 node_865 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36205c5d0 leaf_429 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12b200100 node_866 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12b200100 node_865 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12b200100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12b200100 node_866 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_868 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x324eb8e80 layers.23.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_853 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_869 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_868 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_869 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36205b590 leaf_431 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12b200100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36205c9e0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36205ccf0 leaf_433 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x133200200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_434 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_879 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3269b8f40 layers.23.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_880 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_879 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_851 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_881 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_880 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_882 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_881 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x32bb4d080 layers.23.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_883 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x3272bcfc0 layers.23.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_882 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_884 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_883 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_885 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x32a31d040 layers.23.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_882 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_886 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_884 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_885 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_887 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x328aed000 layers.23.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_886 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_888 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_887 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_880 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_889 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_888 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_890 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_889 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x32df511c0 layers.24.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_891 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x32c451100 layers.24.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_890 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_892 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_891 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_892 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36205e820 leaf_438 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12b9fe100 node_894 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36205ed30 leaf_439 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12b9fe100 node_895 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12b9fe100 node_894 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_896 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x32cd51140 layers.24.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_890 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_897 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_896 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_897 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1336005fe node_899 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36205ef40 leaf_441 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1336005fe node_900 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1336005fe node_899 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x133600200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36205fe80 leaf_446 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12b600100 node_902 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36205f450 leaf_447 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12b600100 node_903 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12b600100 node_902 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12b600100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12b600100 node_903 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_905 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x32bb510c0 layers.24.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_890 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_906 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_905 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_906 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36205e410 leaf_449 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12b600100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36205f860 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36205fb70 leaf_451 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x133600200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_452 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_916 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x32d651180 layers.24.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_917 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_916 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_888 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_918 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_917 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_919 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_918 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x3327e52c0 layers.24.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_920 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x32df55200 layers.24.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_919 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_921 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_920 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_922 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x330fb5280 layers.24.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_919 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_923 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_921 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_922 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_924 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x32f785240 layers.24.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_923 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_925 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_924 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_917 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_926 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_925 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_927 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_926 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x334be9400 layers.25.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_928 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3330e9340 layers.25.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_927 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_929 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_928 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_929 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x3620616a0 leaf_456 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12bdfe100 node_931 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362061bb0 leaf_457 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12bdfe100 node_932 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12bdfe100 node_931 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_933 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3339e9380 layers.25.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_927 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_934 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_933 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_934 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x133a005fe node_936 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362061dc0 leaf_459 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x133a005fe node_937 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x133a005fe node_936 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x133a00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362062d00 leaf_464 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12ba00100 node_939 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620622d0 leaf_465 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12ba00100 node_940 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12ba00100 node_939 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12ba00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12ba00100 node_940 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_942 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3327e9300 layers.25.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_927 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_943 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_942 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_943 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362061290 leaf_467 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12ba00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x3620626e0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620629f0 leaf_469 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x133a00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_470 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_953 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3342e93c0 layers.25.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_954 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_953 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_925 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_955 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_954 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_956 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_955 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x33947d500 layers.25.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_957 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x334bed440 layers.25.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_956 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_958 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_957 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_959 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x337c4d4c0 layers.25.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_956 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_960 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_958 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_959 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_961 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x33641d480 layers.25.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_960 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_962 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_961 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_954 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_963 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_962 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_964 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_963 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x33b881640 layers.26.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_965 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x339d81580 layers.26.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_964 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_966 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_965 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_966 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362064520 leaf_474 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12c1fe100 node_968 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362064a30 leaf_475 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12c1fe100 node_969 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12c1fe100 node_968 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_970 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x33a6815c0 layers.26.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_964 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_971 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_970 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_971 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x133e005fe node_973 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362064c40 leaf_477 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x133e005fe node_974 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x133e005fe node_973 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x133e00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362065b80 leaf_482 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12be00100 node_976 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362065150 leaf_483 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12be00100 node_977 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12be00100 node_976 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12be00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12be00100 node_977 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_979 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x339481540 layers.26.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_964 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_980 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_979 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_980 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362064110 leaf_485 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12be00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362065560 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362065870 leaf_487 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x133e00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_488 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_990 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x33af81600 layers.26.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_991 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_990 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_962 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_992 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_991 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_993 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_992 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x340115740 layers.26.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_994 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x33b885680 layers.26.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_993 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_995 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_994 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_996 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x33e8e5700 layers.26.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_993 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_997 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_995 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_996 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_998 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x33d0b56c0 layers.26.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_997 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_999 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_998 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_991 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1000 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_999 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1001 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1000 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x342519880 layers.27.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_1002 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x340a197c0 layers.27.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1001 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_1003 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_1002 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_1003 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x3620673a0 leaf_492 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12c5fe100 node_1005 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620678b0 leaf_493 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12c5fe100 node_1006 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12c5fe100 node_1005 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1007 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x341319800 layers.27.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1001 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1008 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1007 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1008 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1342005fe node_1010 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362067ac0 leaf_495 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1342005fe node_1011 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1342005fe node_1010 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x134200200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362068a00 leaf_500 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12c200100 node_1013 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362067fd0 leaf_501 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12c200100 node_1014 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12c200100 node_1013 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12c200100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12c200100 node_1014 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_1016 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x340119780 layers.27.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1001 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_1017 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_1016 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_1017 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362066f90 leaf_503 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12c200100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x3620683e0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620686f0 leaf_505 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x134200200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_506 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_1027 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x341c19840 layers.27.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1028 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_1027 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_999 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_1029 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1028 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1030 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_1029 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x346dad980 layers.27.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_1031 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x34251d8c0 layers.27.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1030 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_1032 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_1031 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_1033 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x34557d940 layers.27.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1030 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_1034 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_1032 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_1033 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_1035 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x343d4d900 layers.27.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_1034 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1036 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_1035 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1028 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1037 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1036 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1038 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1037 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x3491b1ac0 layers.28.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_1039 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3476b1a00 layers.28.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1038 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_1040 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_1039 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_1040 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36206a220 leaf_510 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12c9fe100 node_1042 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36206a730 leaf_511 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12c9fe100 node_1043 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12c9fe100 node_1042 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1044 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x347fb1a40 layers.28.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1038 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1045 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1044 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1045 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1346005fe node_1047 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36206a940 leaf_513 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1346005fe node_1048 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1346005fe node_1047 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x134600200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36206b880 leaf_518 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12c600100 node_1050 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36206ae50 leaf_519 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12c600100 node_1051 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12c600100 node_1050 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12c600100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12c600100 node_1051 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_1053 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x346db19c0 layers.28.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1038 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_1054 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_1053 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_1054 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362069e10 leaf_521 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12c600100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36206b260 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36206b570 leaf_523 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x134600200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_524 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_1064 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3488b1a80 layers.28.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1065 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_1064 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1036 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_1066 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1065 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1067 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_1066 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x34da45bc0 layers.28.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_1068 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x3491b5b00 layers.28.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1067 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_1069 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_1068 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_1070 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x34c215b80 layers.28.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1067 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_1071 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_1069 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_1070 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_1072 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x34a9e5b40 layers.28.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_1071 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1073 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_1072 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1065 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1074 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1073 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1075 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1074 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x34fe49d00 layers.29.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_1076 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x34e349c40 layers.29.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1075 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_1077 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_1076 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_1077 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36206d0a0 leaf_528 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12cdfe100 node_1079 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36206d5b0 leaf_529 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12cdfe100 node_1080 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12cdfe100 node_1079 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1081 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x34ec49c80 layers.29.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1075 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1082 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1081 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1082 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x134a005fe node_1084 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36206d7c0 leaf_531 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x134a005fe node_1085 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x134a005fe node_1084 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x134a00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36206e700 leaf_536 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12ca00100 node_1087 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36206dcd0 leaf_537 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12ca00100 node_1088 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12ca00100 node_1087 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12ca00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12ca00100 node_1088 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_1090 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x34da49c00 layers.29.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1075 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_1091 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_1090 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_1091 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36206cc90 leaf_539 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12ca00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x36206e0e0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x36206e3f0 leaf_541 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x134a00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_542 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_1101 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x34f549cc0 layers.29.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1102 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_1101 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1073 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_1103 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1102 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1104 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_1103 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x3546dde00 layers.29.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_1105 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x34fe4dd40 layers.29.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1104 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_1106 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_1105 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_1107 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x352eaddc0 layers.29.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1104 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_1108 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_1106 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_1107 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_1109 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x35167dd80 layers.29.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_1108 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1110 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_1109 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1102 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1111 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1110 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1112 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1111 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x356ae1f40 layers.30.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_1113 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x354fe1e80 layers.30.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1112 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_1114 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_1113 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_1114 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36206ff20 leaf_546 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12d1fe100 node_1116 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362070430 leaf_547 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12d1fe100 node_1117 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12d1fe100 node_1116 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1118 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3558e1ec0 layers.30.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1112 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1119 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1118 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1119 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x134e005fe node_1121 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362070640 leaf_549 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x134e005fe node_1122 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x134e005fe node_1121 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x134e00200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362071580 leaf_554 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12ce00100 node_1124 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362070b50 leaf_555 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12ce00100 node_1125 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12ce00100 node_1124 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12ce00100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12ce00100 node_1125 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_1127 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3546e1e40 layers.30.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1112 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_1128 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_1127 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_1128 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x36206fb10 leaf_557 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12ce00100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362070f60 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362071270 leaf_559 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x134e00200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_560 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_1138 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x3561e1f00 layers.30.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1139 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_1138 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1110 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_1140 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1139 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1141 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_1140 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x35b376040 layers.30.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_1142 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x356ae5f80 layers.30.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1141 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_1143 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_1142 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_1144 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x359b46000 layers.30.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1141 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_1145 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_1143 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_1144 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_1146 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x358315fc0 layers.30.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_1145 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1147 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_1146 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1139 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1148 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1147 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1149 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1148 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x35d77a180 layers.31.attention_norm.weight | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_1150 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x35bc7a0c0 layers.31.attention.wk.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1149 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_1151 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b0000 node_1150 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 node_1151 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362072da0 leaf_564 | |
DST f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12d5fe100 node_1153 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620732b0 leaf_565 | |
DST f16 CPY 1 4096 1 1 1 2 8192 8192 8192 1 0x12d5fe100 node_1154 | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051b0000 Kcur | |
SRC1 f16 VIEW 1 4096 1 1 1 2 8192 8192 8192 1 0x12d5fe100 node_1153 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1155 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x35c57a100 layers.31.attention.wv.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1149 | |
DST f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1156 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1155 | |
DST f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC0 f32 RESHAPE 2 4096 1 1 1 4 16384 16384 16384 1 0x1051b4000 node_1156 | |
DST f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1352005fe node_1158 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620734c0 leaf_567 | |
DST f16 CPY 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1352005fe node_1159 | |
SRC0 f32 TRANSPOSE 2 1 4096 1 1 16384 4 16384 16384 1 0x1051b4000 Vcur | |
SRC1 f16 VIEW 2 1 4096 1 1 2 1024 4194304 4194304 1 0x1352005fe node_1158 | |
DST f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x135200200 V | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x12d600200 cache_v | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x362074400 leaf_572 | |
DST f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12d200100 node_1161 | |
SRC0 f16 NONE 1 67108864 1 1 1 2 134217728 134217728 134217728 0 0x125600100 cache_k | |
OPT i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620739d0 leaf_573 | |
DST f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12d200100 node_1162 | |
SRC0 f16 VIEW 1 2097152 1 1 1 2 4194304 4194304 4194304 1 0x12d200100 node_1161 | |
DST f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12d200100 K | |
SRC0 f16 RESHAPE 3 128 32 512 1 2 256 8192 4194304 1 0x12d200100 node_1162 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_1164 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x35b37a080 layers.31.attention.wq.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1149 | |
DST f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_1165 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051ac000 node_1164 | |
DST f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
SRC0 f32 RESHAPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 node_1165 | |
SRC1 i32 NONE 1 3 1 1 1 4 12 12 12 0 0x362072990 leaf_575 | |
DST f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
SRC0 f32 ROPE 3 128 32 1 1 4 512 16384 16384 1 0x1051ac000 Qcur | |
DST f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC0 f16 PERMUTE 3 128 512 32 1 2 8192 256 4194304 1 0x12d200100 K | |
SRC1 f32 PERMUTE 3 128 1 32 1 4 16384 512 16384 1 0x1051ac000 Q | |
DST f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC0 f32 MUL_MAT 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ | |
SRC1 f32 NONE 1 1 1 1 1 4 4 4 4 0 0x362073de0 1/sqrt(n_embd/n_head) | |
DST f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
SRC0 f32 SCALE 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_scaled | |
SRC1 i32 NONE 1 2 1 1 1 4 8 8 8 0 0x3620740f0 leaf_577 | |
DST f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
SRC0 f32 DIAG_MASK_INF 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_masked | |
DST f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
SRC0 f16 VIEW 3 512 128 32 1 2 1024 131072 4194304 1 0x135200200 V | |
SRC1 f32 SOFT_MAX 3 512 1 32 1 4 2048 2048 65536 1 0x1051b8000 KQ_soft_max | |
DST f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC0 f32 MUL_MAT 3 128 1 32 1 4 512 512 16384 1 0x1051c8000 KQV | |
DST f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
SRC0 f32 PERMUTE 3 128 32 1 1 4 512 512 16384 1 0x1051c8000 KQV_merged | |
SRC1 f32 NONE 2 4096 1 1 1 4 16384 16384 16384 0 0x1051cc000 leaf_578 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_1175 | |
SRC0 q4_0 NONE 2 4096 4096 1 1 18 2304 9437184 9437184 0 0x35ce7a140 layers.31.attention.wo.weight | |
SRC1 f32 CPY 2 4096 1 1 1 4 16384 16384 16384 1 0x1051cc000 KQV_merged_contiguous | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1176 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x1051d0000 node_1175 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1147 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_1177 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1176 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1178 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x392018000 node_1177 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x36200e280 layers.31.ffn_norm.weight | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_1179 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x35d77e1c0 layers.31.feed_forward.w1.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1178 | |
DST f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_1180 | |
SRC0 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x39202ac00 node_1179 | |
DST f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_1181 | |
SRC0 q4_0 NONE 2 4096 11008 1 1 18 2304 25362432 25362432 0 0x3607de240 layers.31.feed_forward.w3.weigh | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x39201c000 node_1178 | |
DST f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_1182 | |
SRC0 f32 SILU 2 11008 1 1 1 4 44032 44032 44032 1 0x392035800 node_1180 | |
SRC1 f32 MUL_MAT 2 11008 1 1 1 4 44032 44032 44032 1 0x392020000 node_1181 | |
DST f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_1183 | |
SRC0 q4_0 NONE 2 11008 4096 1 1 18 6192 25362432 25362432 0 0x35efae200 layers.31.feed_forward.w2.weigh | |
SRC1 f32 MUL 2 11008 1 1 1 4 44032 44032 44032 1 0x392040400 node_1182 | |
DST f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1184 | |
SRC0 f32 MUL_MAT 2 4096 1 1 1 4 16384 16384 16384 1 0x39204b000 node_1183 | |
SRC1 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x392014000 node_1176 | |
DST f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1185 | |
SRC0 f32 ADD 2 4096 1 1 1 4 16384 16384 16384 1 0x39204f000 node_1184 | |
DST f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1186 | |
SRC0 f32 RMS_NORM 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a4000 node_1185 | |
SRC1 f32 NONE 1 4096 1 1 1 4 16384 16384 16384 0 0x2846b9a40 norm.weight | |
DST f32 MUL_MAT 2 32000 1 1 1 4 128000 128000 128000 1 0x1051ac000 node_1187 | |
SRC0 q4_0 NONE 2 4096 32000 1 1 18 2304 73728000 73728000 0 0x2846bda80 output.weight | |
SRC1 f32 MUL 2 4096 1 1 1 4 16384 16384 16384 1 0x1051a8000 node_1186 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment