poc teacher notebook

Author: paramkpr

.gitignore

@@ -28,7 +28,8 @@ venv312/
 venv311/
 # Jupyter Notebook
 .ipynb_checkpoints
-
+tmp/
+tmp*/
 # Distribution / packaging
 .Python
 env/

notebooks/02_baseline_models.ipynb

@@ -0,0 +1,184 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Training the Teacher Model\n",
+    "The first step in the pipeline is to train a teacher model on the SST-2 dataset. This model will be used to classify the synthetic data generated by the generator.  \n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "0389b731c3af4704bc3f0564bd5f6417",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Map:   0%|          | 0/256 [00:00<?, ? examples/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "import torch\n",
+    "from datasets import load_dataset\n",
+    "from transformers import AutoModelForSequenceClassification, AutoTokenizer, DataCollatorWithPadding\n",
+    "\n",
+    "datasets = load_dataset(\"glue\", \"sst2\", split=\"train[:256]\")\n",
+    "datasets = datasets.rename_column(\"label\", \"labels\")\n",
+    "\n",
+    "tokenizer = AutoTokenizer.from_pretrained(\"prajjwal1/bert-small\", use_fast=True)\n",
+    "def tokenize_function(examples):\n",
+    "    return tokenizer(examples[\"sentence\"], truncation=True, max_length=32)\n",
+    "\n",
+    "tokenized_datasets = datasets.map(tokenize_function, batched=True)\n",
+    "tokenized_datasets.set_format(type=\"torch\", columns=[\"input_ids\", \"attention_mask\", \"labels\"])\n",
+    "\n",
+    "\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Some weights of BertForSequenceClassification were not initialized from the model checkpoint at prajjwal1/bert-small and are newly initialized: ['classifier.bias', 'classifier.weight']\n",
+      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Model loaded on mps\n"
+     ]
+    }
+   ],
+   "source": [
+    "model = AutoModelForSequenceClassification.from_pretrained(\"prajjwal1/bert-small\")\n",
+    "device = \"cuda\" if torch.cuda.is_available() else \"mps\" if torch.backends.mps.is_available() else \"cpu\"\n",
+    "model.to(device)\n",
+    "print(f\"Model loaded on {device}\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 26,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from transformers import Trainer, TrainingArguments\n",
+    "\n",
+    "args = TrainingArguments(\n",
+    "    output_dir=\"tmp/teacher_poc\",\n",
+    "    per_device_train_batch_size=2,\n",
+    "    per_device_eval_batch_size=2,\n",
+    "    num_train_epochs=1,\n",
+    "    learning_rate=2e-5,\n",
+    "    eval_strategy=\"steps\",\n",
+    "    eval_steps=50,\n",
+    ")\n",
+    "\n",
+    "trainer = Trainer(\n",
+    "    model=model,\n",
+    "    args=args,\n",
+    "    train_dataset=tokenized_datasets,\n",
+    "    eval_dataset=tokenized_datasets.shuffle(seed=0).select(range(64)),  # tiny eval slice\n",
+    "    data_collator=DataCollatorWithPadding(tokenizer),\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 27,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "\n",
+       "    <div>\n",
+       "      \n",
+       "      <progress value='128' max='128' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
+       "      [128/128 00:18, Epoch 1/1]\n",
+       "    </div>\n",
+       "    <table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       " <tr style=\"text-align: left;\">\n",
+       "      <th>Step</th>\n",
+       "      <th>Training Loss</th>\n",
+       "      <th>Validation Loss</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <td>50</td>\n",
+       "      <td>No log</td>\n",
+       "      <td>0.669309</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <td>100</td>\n",
+       "      <td>No log</td>\n",
+       "      <td>0.646186</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table><p>"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/plain": [
+       "TrainOutput(global_step=128, training_loss=0.6919310092926025, metrics={'train_runtime': 24.1107, 'train_samples_per_second': 10.618, 'train_steps_per_second': 5.309, 'total_flos': 349921897560.0, 'train_loss': 0.6919310092926025, 'epoch': 1.0})"
+      ]
+     },
+     "execution_count": 27,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "trainer.train()"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python3.11 (sentisynth)",
+   "language": "python",
+   "name": "auctionn"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.2"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}