softmax.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "c2a2be05-6c9d-4685-bcf1-f2c6a61410f7",
   "metadata": {},
   "source": [
    "# Testing Softmax for accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "93ad8ecb-2c35-456a-a741-287d8151b9ee",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Importing torch...\n",
      "Importing HF...\n",
      "Importing python modules...\n",
      "Done!\n"
     ]
    }
   ],
   "source": [
    "print(\"Importing torch...\")\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "print(\"Importing HF...\")\n",
    "from transformers import AutoTokenizer, AutoModelForCausalLM\n",
    "print(\"Importing python modules...\")\n",
    "from timehelp import time_start, time_end\n",
    "from model_wrapper import Model, ModelFamily, MultipleChoiceStrategy\n",
    "import re\n",
    "print(\"Done!\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "689b042d-543a-4b65-8c67-854478aa1a84",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[2024-05-06@16:58:57|model.device] Starting timer.\n",
      "Configuring torch device...\n",
      "Using device: cuda:0 aka cuda:0\n",
      "[2024-05-06@16:58:57|model.device] Time elapsed: 23ms\n",
      "[2024-05-06@16:58:57|model.tokenizer] Starting timer.\n",
      "[2024-05-06@16:58:57|model.tokenizer] Time elapsed: 209ms\n",
      "[2024-05-06@16:58:57|model.model] Starting timer.\n",
      "Obtaining model...\n",
      "[2024-05-06@16:59:01|model.model] Time elapsed: 3s 364ms\n"
     ]
    }
   ],
   "source": [
    "model = Model(ModelFamily.CodeGen1.mono[\"350M\"])\n",
    "model.configure(time=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b81192c8-c550-4ab8-b127-4eb419fe8ba1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[2024-05-06@17:00:01|model.tokenize] Starting timer.\n",
      "[2024-05-06@17:00:01|model.tokenize] Time elapsed: 106ms\n",
      "Token count in input: 4\n"
     ]
    }
   ],
   "source": [
    "prompt = \"\"\"for i in range\"\"\"\n",
    "tokens = model.tokenize(prompt, time=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fdfcc101-6f97-457f-ae6d-40324383346e",
   "metadata": {},
   "source": [
    "## Sort by logits"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "19ffb1ea-4938-494e-9928-bbb28a2c314f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\"(\"\tlogits=27.749189376831055\ttoken=7\n",
      "\" (\"\tlogits=23.678096771240234\ttoken=357\n",
      "\"(-\"\tlogits=21.390438079833984\ttoken=32590\n",
      "\"((\"\tlogits=20.967388153076172\ttoken=19510\n",
      "\"(_\"\tlogits=19.828998565673828\ttoken=28264\n",
      "\"():\"\tlogits=18.87956428527832\ttoken=33529\n",
      "\":\"\tlogits=18.829797744750977\ttoken=25\n"
     ]
    }
   ],
   "source": [
    "base_output = model.model_no_grad(input_ids=tokens[\"input_ids\"])\n",
    "base_logits = base_output.logits[:, -1, :]\n",
    "\n",
    "logits_cpu = base_logits.cpu().detach().numpy().flatten()\n",
    "indices_sorted = torch.argsort(base_logits).cpu().numpy().flatten()\n",
    "sorted_pos = [(logits_cpu[idx], idx) for idx in range(len(indices_sorted))]\n",
    "sorted_pos.sort(key=lambda x: x[0], reverse=True)\n",
    "\n",
    "display_info = [\n",
    "    [ logits, index, f\"\\\"{model.decode([ index ])}\\\"\" ]\n",
    "    for logits, index\n",
    "    in sorted_pos[:7]\n",
    "]\n",
    "\n",
    "print(\"\\n\".join(\n",
    "    f\"{rep}\\tlogits={logits}\\ttoken={index}\"\n",
    "    for logits, index, rep\n",
    "    in display_info\n",
    "))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "064c1ff4-93d9-4685-ac57-310eb89b51ab",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "def graph(display_info, kind):\n",
    "    x_axis = [ rep for logits, index, rep in display_info ]\n",
    "    y_axis = [ logits for logits, index, rep in display_info ]\n",
    "    plt.figure(figsize=(10, 6))\n",
    "    plt.bar(x_axis, y_axis, color=\"skyblue\")\n",
    "    plt.xlabel(\"Decoded Tokens\")\n",
    "    plt.ylabel(kind)\n",
    "    plt.title(f\"Top Tokens and {kind}\")\n",
    "    # plt.xticks(rotation=45, ha=\"right\")  # Rotate x-axis labels for better readability\n",
    "    for tick in plt.gca().get_xticklabels():\n",
    "        tick.set_fontname(\"monospace\")\n",
    "    plt.tight_layout()  # Adjust layout to prevent label cutoffs\n",
    "    plt.show()\n",
    "\n",
    "graph(display_info, \"Logits\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "988e1703-4f60-400d-9d8b-5af86092a4f3",
   "metadata": {},
   "source": [
    "## Sort by softmax"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "651bc437-c303-4b2e-83ca-a72581bbd5a0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\"(\"\tPr=97.92%\ttoken=7\n",
      "\" (\"\tPr=1.67%\ttoken=357\n",
      "\"(-\"\tPr=0.16%\ttoken=32590\n",
      "\"((\"\tPr=0.11%\ttoken=19510\n",
      "\"(_\"\tPr=0.03%\ttoken=28264\n",
      "\"():\"\tPr=0.01%\ttoken=33529\n",
      "\":\"\tPr=0.01%\ttoken=25\n"
     ]
    }
   ],
   "source": [
    "softmax = torch.nn.Softmax(dim=-1)\n",
    "\n",
    "distribution = softmax(base_logits)\n",
    "distribution\n",
    "\n",
    "distribution_cpu = distribution.cpu().detach().numpy().flatten()\n",
    "indices_sorted = torch.argsort(distribution).cpu().numpy().flatten()\n",
    "sorted_pos = [(distribution_cpu[idx], idx) for idx in range(len(indices_sorted))]\n",
    "sorted_pos.sort(key=lambda x: x[0], reverse=True)\n",
    "\n",
    "display_info = [\n",
    "    [ pr, index, f\"\\\"{model.decode([ index ])}\\\"\" ]\n",
    "    for pr, index\n",
    "    in sorted_pos[:7]\n",
    "]\n",
    "\n",
    "print(\"\\n\".join(\n",
    "    f\"{rep}\\tPr={int(pr * 10000) / 100}%\\ttoken={index}\"\n",
    "    for pr, index, rep\n",
    "    in display_info\n",
    "))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "8b996457-e123-43ac-8227-636b6174f310",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "graph(display_info, \"Probability\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}