| $ conda activate ml | |
| (ml) $ conda install ipykernel | |
| (ml) $ ipython kernel install --user --name=<any_name_for_kernel> | |
| (ml) $ conda deactivate |
| import torchaudio | |
| def eval(audio, text): | |
| # convert audio to 16000 sample rate | |
| audio = torchaudio.transforms.Resample(orig_freq=44100, new_freq=16000)(torch.tensor(audio).unsqueeze(0)).squeeze() | |
| # process text | |
| tokenized_seq = torch.tensor([processor.tokenizer(text, add_special_tokens=True).input_ids]).to(device) | |
| decoder_input_ids = tokenized_seq[:, 1:] | |
| decoder_input_ids_right_shifted = tokenized_seq[:, :-1] | |
| # process audio |
| word | jyutping | |
|---|---|---|
| 一剎那 | no4 | |
| 七竅 | hiu3 | |
| 七鰓鰻 | soi1 | |
| 丈母娘 | noeng4 | |
| 乓 | bam1 | |
| 九層塔 | taap3 | |
| 乞兒 | haat1 | |
| 乾坤 | kin4 | |
| 乾闥婆 | taat3 |
| const resampling = (audioBuffer: AudioBuffer, targetSampleRate: number): Promise<AudioBuffer> => { | |
| const offlineAudioContext = new OfflineAudioContext(1, audioBuffer.length, targetSampleRate); | |
| const source = offlineAudioContext.createBufferSource(); | |
| source.buffer = audioBuffer; | |
| source.connect(offlineAudioContext.destination); | |
| source.start(); | |
| return offlineAudioContext.startRendering(); |
| You are asked to come up with a set of 20 diverse and detailed roleplay scenarios. These scenarios will be given to a large language model and we will evaluate the large language model for completing the roleplay effectively. | |
| Here are the requirements: | |
| 1. Ensure the roles and scenarios are diverse, covering different contexts, professions, and situations. | |
| 2. The language used for the scenarios should also be diverse. For example, include both formal and informal dialogues. | |
| 3. The type of scenarios should be diverse. Include diverse types of interactions like customer service, medical consultations, casual conversations, educational settings, etc. | |
| 4. A large language model should be able to complete the roleplay. For example, do not include scenarios that require real-time actions or physical interactions. | |
| 5. The scenarios and dialogues should be in **Cantonese**. | |
| 6. Each scenario should be detailed, providing a clear context and background. Include relevant information such as the setting, characters involv |
| adb -e shell date $(date +%m%d%H%M%Y.%S) |
Here is some thoughts of mine on the vision of AI Smart Watch. I hope you enjoy it.
Running LLM(Large language model) on a embedding device is a trendy topic in the field of AI. The AI Smart Watch is a typical example of such devices. It is also a challenging task because of the limited resources of the device. In this article, I will explore the vision and the solution of AI Smart Watch.
| User:講一個關於維尼大大嘅故仔畀我聽 | |
| CantoneseLLM: 維尼大大係一個好聰明、好有創意嘅人。佢成日都會諗到啲好有趣嘅主意。 | |
| 有一次，維尼大大去咗公園玩。佢見到好多細路仔喺度玩滑梯。佢突然間諗到，如果將滑梯改造成一個旋轉木馬，一定會好受小朋友歡迎。 | |
| 於是，維尼大大就開始動手整旋轉木馬。佢用咗好多唔同嘅材料嚟整，包括木頭、金屬同埋塑膠。佢仲用咗好多唔同嘅顏色嚟裝飾旋轉木馬。 | |
| 終於，維尼大大整好咗旋轉木馬。佢將旋轉木馬放喺公園入面，細路仔一見到旋轉木馬就即刻衝埋去玩。旋轉木馬好受小朋友歡迎，好多細路仔都排隊玩。 |
| { | |
| "assessment": { | |
| "title": "中文運用模擬試卷", | |
| "access": "public", | |
| "type": "assessment", | |
| "model": "assessmentSnapshot", | |
| "description": "", | |
| "isTimed": true, | |
| "timerFor": "assessmentTimed", | |
| "onTimerExpiration": "autoSubmit", |
| const wait = ms => new Promise(resolve => setTimeout(resolve, ms)); | |
| const timeout = (p, ms) => Promise.race([p, wait(ms).then(() => { | |
| throw new Error("Timeout after " + ms + " ms"); | |
| })]); | |
| export const pubsubFunction: CloudEventFunction<{ | |
| message: { data: string }; | |
| }> = async () => { | |
| timeout(checkShellTransactions(), 3 * 60 * 1000) // 3 minutes ms | |
| }; |
