Skip to content

Instantly share code, notes, and snippets.

@aboSamoor

aboSamoor/gist:7161552

Created October 25, 2013 20:46
Show Gist options
  • Save aboSamoor/7161552 to your computer and use it in GitHub Desktop.
Save aboSamoor/7161552 to your computer and use it in GitHub Desktop.
Download ZIP
Raw
gistfile1.txt
{
"metadata": {
"name": ""
},
"nbformat": 3,
"nbformat_minor": 0,
"worksheets": [
{
"cells": [
{
"cell_type": "code",
"collapsed": false,
"input": [
"cd ./../"
],
"language": "python",
"metadata": {},
"outputs": [
{
"output_type": "stream",
"stream": "stdout",
"text": [
"/media/jarra/pdp/PDP-DL\n"
]
}
],
"prompt_number": 1
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"%pylab inline\n",
"from pdp import run_pdp, PDPModel\n",
"from collections import OrderedDict\n",
"import pdp_utils\n",
"# pdp model, identity representation\n",
"pdp_model = PDPModel()\n",
"pdp_filename = 'resources/spieces_toy.rel'\n",
"pdp_model.load_pdp_data(pdp_filename)\n",
"pdp_model.n_rep=1\n",
"pdp_model.n_hidden=8\n",
"pdp_model.rep_activation=pdp_utils.identity\n",
"losses = []\n",
"snapshots = OrderedDict()\n",
"pdp_model.set_adaptive_lr(0.5)\n",
"losses, snapshots = run_pdp(pdp_model, num_iter=1000, snapshot_freq=1)\n",
"\n",
"pdp_model.set_adaptive_lr(0.5)\n",
"pdp_model.init_G()\n",
"losses, snapshots = run_pdp(pdp_model, num_iter=5000, snapshot_freq=1,\n",
" continuation=[losses, snapshots])\n",
"it = range(1050, len(losses))\n",
"plot(it, array(losses)[it])\n",
"xscale('log')\n",
"ylim((0, max(array(losses)[it])))\n",
"show()"
],
"language": "python",
"metadata": {},
"outputs": [
{
"output_type": "stream",
"stream": "stdout",
"text": [
"Populating the interactive namespace from numpy and matplotlib\n"
]
},
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAXkAAAEBCAYAAACdctWRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
"text": [
"<matplotlib.figure.Figure at 0xb861b50>"
]
}
],
"prompt_number": 8
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"%pylab inline\n",
"from pdp import run_pdp, PDPModel\n",
"from collections import OrderedDict\n",
"import pdp_utils\n",
"# pdp model, identity representation\n",
"pdp_model = PDPModel()\n",
"pdp_filename = 'resources/spieces_toy.rel'\n",
"pdp_model.load_pdp_data(pdp_filename)\n",
"pdp_model.n_rep=1\n",
"pdp_model.n_hidden=8\n",
"pdp_model.rep_activation=pdp_utils.identity\n",
"losses = []\n",
"snapshots = OrderedDict()\n",
"pdp_model.set_adaptive_lr(0.5)\n",
"losses, snapshots = run_pdp(pdp_model, num_iter=6000, snapshot_freq=1)\n",
"\n",
"#pdp_model.set_adaptive_lr(0.5)\n",
"#pdp_model.init_G()\n",
"#losses, snapshots = run_pdp(pdp_model, num_iter=5000, snapshot_freq=1,\n",
"# continuation=[losses, snapshots])\n",
"it = range(1050, len(losses))\n",
"plot(it, array(losses)[it])\n",
"xscale('log')\n",
"ylim((0, max(array(losses)[it])))\n",
"show()"
],
"language": "python",
"metadata": {},
"outputs": [
{
"output_type": "stream",
"stream": "stdout",
"text": [
"Populating the interactive namespace from numpy and matplotlib\n"
]
},
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAXAAAAEBCAYAAABhZ/5bAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
"text": [
"<matplotlib.figure.Figure at 0x779c050>"
]
}
],
"prompt_number": 9
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"%pylab inline\n",
"from pdp import run_pdp, PDPModel\n",
"from collections import OrderedDict\n",
"import pdp_utils\n",
"# pdp model, identity representation\n",
"pdp_model = PDPModel()\n",
"pdp_filename = 'resources/spieces_toy.rel'\n",
"pdp_model.load_pdp_data(pdp_filename)\n",
"pdp_model.n_rep=1\n",
"pdp_model.n_hidden=8\n",
"pdp_model.rep_activation=pdp_utils.identity\n",
"losses = []\n",
"snapshots = OrderedDict()\n",
"pdp_model.set_decreasing_adaptive_lr(0.5)\n",
"losses, snapshots = run_pdp(pdp_model, num_iter=6000, snapshot_freq=1)\n",
"\n",
"#pdp_model.set_adaptive_lr(0.5)\n",
"#pdp_model.init_G()\n",
"#losses, snapshots = run_pdp(pdp_model, num_iter=5000, snapshot_freq=1,\n",
"# continuation=[losses, snapshots])\n",
"it = range(1050, len(losses))\n",
"plot(it, array(losses)[it])\n",
"xscale('log')\n",
"ylim((0, max(array(losses)[it])))\n",
"show()"
],
"language": "python",
"metadata": {},
"outputs": [
{
"output_type": "stream",
"stream": "stdout",
"text": [
"Populating the interactive namespace from numpy and matplotlib\n"
]
},
{
"metadata": {},
"output_type": "display_data",
"png": "iVBORw0KGgoAAAANSUhEUgAAAXAAAAEBCAYAAABhZ/5bAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
"text": [
"<matplotlib.figure.Figure at 0x508d550>"
]
}
],
"prompt_number": 10
}
],
"metadata": {}
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment