fix: repair module 2 notebook execution

.gitignore

@@ -13,6 +13,7 @@ src/models/saved_emotion_model/
 src/models/saved_*/
 checkpoints/
 runs/
+.hf_cache/
 
 # Local secrets and editor files
 .env

notebooks/module_2_emotion_training.ipynb

@@ -29,7 +29,13 @@
       "metadata": {},
       "outputs": [],
       "source": [
-        "!nvidia-smi"
+        "import shutil\n",
+        "import subprocess\n",
+        "\n",
+        "if shutil.which('nvidia-smi'):\n",
+        "    subprocess.run(['nvidia-smi'], check=False)\n",
+        "else:\n",
+        "    print('No NVIDIA GPU detected in this runtime. Use Colab T4 for full training.')"
       ]
     },
     {
@@ -39,7 +45,27 @@
       "metadata": {},
       "outputs": [],
       "source": [
-        "%pip -q install -U transformers datasets accelerate evaluate scikit-learn pandas"
+        "import os\n",
+        "import subprocess\n",
+        "import sys\n",
+        "\n",
+        "if os.getenv('FAST_DEV_RUN', '0') == '1':\n",
+        "    print('Skipping package install during local FAST_DEV_RUN.')\n",
+        "else:\n",
+        "    subprocess.check_call([\n",
+        "        sys.executable,\n",
+        "        '-m',\n",
+        "        'pip',\n",
+        "        'install',\n",
+        "        '-q',\n",
+        "        '-U',\n",
+        "        'transformers>=4.40,<5',\n",
+        "        'datasets>=2.18,<5',\n",
+        "        'accelerate>=0.28',\n",
+        "        'evaluate>=0.4',\n",
+        "        'scikit-learn>=1.4',\n",
+        "    ])\n",
+        "# Do not upgrade pandas in Colab; Colab/GPU packages often pin pandas < 3."
       ]
     },
     {
@@ -72,6 +98,7 @@
       "outputs": [],
       "source": [
         "import json\n",
+        "import os\n",
         "import sys\n",
         "from inspect import signature\n",
         "from pathlib import Path\n",
@@ -80,16 +107,20 @@
         "import pandas as pd\n",
         "from datasets import load_dataset\n",
         "from sklearn.metrics import accuracy_score, classification_report, confusion_matrix, f1_score\n",
-        "from transformers import AutoModelForSequenceClassification, AutoTokenizer, DataCollatorWithPadding, Trainer, TrainingArguments\n",
+        "from transformers import AutoConfig, AutoModelForSequenceClassification, AutoTokenizer, DataCollatorWithPadding, Trainer, TrainingArguments\n",
         "\n",
         "PROJECT_ROOT = Path.cwd().resolve().parent if Path.cwd().name == 'notebooks' else Path.cwd().resolve()\n",
         "MODEL_DIR = PROJECT_ROOT / 'src' / 'models' / 'saved_emotion_model'\n",
         "REPORT_DIR = PROJECT_ROOT / 'reports' / 'module_2_emotion_classification'\n",
+        "CACHE_DIR = PROJECT_ROOT / '.hf_cache'\n",
         "MODEL_NAME = 'distilbert-base-uncased'\n",
+        "FAST_DEV_RUN = os.getenv('FAST_DEV_RUN', '0') == '1'\n",
         "\n",
         "MODEL_DIR.mkdir(parents=True, exist_ok=True)\n",
         "REPORT_DIR.mkdir(parents=True, exist_ok=True)\n",
-        "sys.path.append(str(PROJECT_ROOT / 'src' / 'models'))"
+        "CACHE_DIR.mkdir(parents=True, exist_ok=True)\n",
+        "sys.path.append(str(PROJECT_ROOT / 'src' / 'models'))\n",
+        "print(f'FAST_DEV_RUN={FAST_DEV_RUN}')"
       ]
     },
     {
@@ -100,6 +131,13 @@
       "outputs": [],
       "source": [
         "dataset = load_dataset('dair-ai/emotion', 'split')\n",
+        "\n",
+        "if FAST_DEV_RUN:\n",
+        "    dataset = dataset.shuffle(seed=42)\n",
+        "    dataset['train'] = dataset['train'].select(range(64))\n",
+        "    dataset['validation'] = dataset['validation'].select(range(32))\n",
+        "    dataset['test'] = dataset['test'].select(range(32))\n",
+        "\n",
         "label_names = dataset['train'].features['label'].names\n",
         "id2label = {i: label for i, label in enumerate(label_names)}\n",
         "label2id = {label: i for i, label in id2label.items()}\n",
@@ -116,14 +154,14 @@
       "metadata": {},
       "outputs": [],
       "source": [
-        "tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)\n",
+        "tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, cache_dir=CACHE_DIR)\n",
         "\n",
         "def tokenize(batch):\n",
         "    return tokenizer(batch['text'], truncation=True, max_length=128)\n",
         "\n",
         "encoded = dataset.map(tokenize, batched=True)\n",
         "encoded = encoded.rename_column('label', 'labels')\n",
-        "encoded.set_format(type='torch', columns=['input_ids', 'attention_mask', 'labels'])\n",
+        "encoded = encoded.remove_columns(['text'])\n",
         "data_collator = DataCollatorWithPadding(tokenizer=tokenizer)"
       ]
     },
@@ -134,13 +172,27 @@
       "metadata": {},
       "outputs": [],
       "source": [
-        "model = AutoModelForSequenceClassification.from_pretrained(\n",
+        "config = AutoConfig.from_pretrained(\n",
         "    MODEL_NAME,\n",
         "    num_labels=len(label_names),\n",
         "    id2label=id2label,\n",
         "    label2id=label2id,\n",
+        "    cache_dir=CACHE_DIR,\n",
         ")\n",
         "\n",
+        "if FAST_DEV_RUN:\n",
+        "    config.dim = 64\n",
+        "    config.hidden_dim = 128\n",
+        "    config.n_layers = 1\n",
+        "    config.n_heads = 2\n",
+        "    model = AutoModelForSequenceClassification.from_config(config)\n",
+        "else:\n",
+        "    model = AutoModelForSequenceClassification.from_pretrained(\n",
+        "        MODEL_NAME,\n",
+        "        config=config,\n",
+        "        cache_dir=CACHE_DIR,\n",
+        "    )\n",
+        "\n",
         "def compute_metrics(eval_pred):\n",
         "    logits, labels = eval_pred\n",
         "    predictions = np.argmax(logits, axis=-1)\n",
@@ -154,13 +206,14 @@
         "    'learning_rate': 2e-5,\n",
         "    'per_device_train_batch_size': 16,\n",
         "    'per_device_eval_batch_size': 32,\n",
-        "    'num_train_epochs': 3,\n",
+        "    'num_train_epochs': 1 if FAST_DEV_RUN else 3,\n",
         "    'weight_decay': 0.01,\n",
-        "    'save_strategy': 'epoch',\n",
-        "    'load_best_model_at_end': True,\n",
+        "    'save_strategy': 'no' if FAST_DEV_RUN else 'epoch',\n",
+        "    'load_best_model_at_end': False if FAST_DEV_RUN else True,\n",
         "    'metric_for_best_model': 'macro_f1',\n",
         "    'greater_is_better': True,\n",
-        "    'logging_steps': 50,\n",
+        "    'logging_steps': 1 if FAST_DEV_RUN else 50,\n",
+        "    'max_steps': 2 if FAST_DEV_RUN else -1,\n",
         "    'report_to': 'none',\n",
         "}\n",
         "\n",
@@ -213,9 +266,10 @@
         "model.save_pretrained(MODEL_DIR)\n",
         "tokenizer.save_pretrained(MODEL_DIR)\n",
         "\n",
-        "report_text = classification_report(test_labels, test_predictions, target_names=label_names, zero_division=0)\n",
-        "report_dict = classification_report(test_labels, test_predictions, target_names=label_names, output_dict=True, zero_division=0)\n",
-        "matrix = confusion_matrix(test_labels, test_predictions)\n",
+        "all_label_ids = list(range(len(label_names)))\n",
+        "report_text = classification_report(test_labels, test_predictions, labels=all_label_ids, target_names=label_names, zero_division=0)\n",
+        "report_dict = classification_report(test_labels, test_predictions, labels=all_label_ids, target_names=label_names, output_dict=True, zero_division=0)\n",
+        "matrix = confusion_matrix(test_labels, test_predictions, labels=all_label_ids)\n",
         "\n",
         "(REPORT_DIR / 'test_classification_report.txt').write_text(report_text, encoding='utf-8')\n",
         "pd.DataFrame(report_dict).transpose().to_csv(REPORT_DIR / 'test_classification_report.csv')\n",