File size: 7,985 Bytes
4089b4a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 |
"""
Utility functions for training and evaluation
"""
import numpy as np
from sklearn.metrics import (
accuracy_score,
precision_recall_fscore_support,
confusion_matrix,
classification_report
)
import matplotlib.pyplot as plt
import seaborn as sns
from typing import Dict, Tuple, List, Optional
import os
def compute_metrics(eval_pred, id2label: Optional[Dict[int, str]] = None) -> Dict[str, float]:
"""
Compute comprehensive metrics for evaluation.
Args:
eval_pred: Tuple of (predictions, labels)
id2label: Optional mapping from label IDs to label names for per-class metrics
Returns:
Dictionary of metrics including overall and per-class metrics
"""
predictions, labels = eval_pred
predictions = np.argmax(predictions, axis=1)
# Overall metrics
accuracy = accuracy_score(labels, predictions)
# Weighted metrics (accounts for class imbalance)
precision_weighted, recall_weighted, f1_weighted, _ = precision_recall_fscore_support(
labels,
predictions,
average='weighted',
zero_division=0
)
# Macro-averaged metrics (treats all classes equally)
precision_macro, recall_macro, f1_macro, _ = precision_recall_fscore_support(
labels,
predictions,
average='macro',
zero_division=0
)
# Micro-averaged metrics (aggregates contributions of all classes)
precision_micro, recall_micro, f1_micro, _ = precision_recall_fscore_support(
labels,
predictions,
average='micro',
zero_division=0
)
metrics = {
'accuracy': accuracy,
'precision_weighted': precision_weighted,
'recall_weighted': recall_weighted,
'f1_weighted': f1_weighted,
'precision_macro': precision_macro,
'recall_macro': recall_macro,
'f1_macro': f1_macro,
'precision_micro': precision_micro,
'recall_micro': recall_micro,
'f1_micro': f1_micro,
}
# Per-class metrics if label mapping is provided
if id2label is not None:
num_classes = len(id2label)
precision_per_class, recall_per_class, f1_per_class, support = precision_recall_fscore_support(
labels,
predictions,
labels=list(range(num_classes)),
average=None,
zero_division=0
)
for i in range(num_classes):
label_name = id2label[i]
metrics[f'precision_{label_name}'] = float(precision_per_class[i])
metrics[f'recall_{label_name}'] = float(recall_per_class[i])
metrics[f'f1_{label_name}'] = float(f1_per_class[i])
metrics[f'support_{label_name}'] = int(support[i])
return metrics
def compute_metrics_factory(id2label: Optional[Dict[int, str]] = None):
"""
Factory function to create compute_metrics with label mapping.
Args:
id2label: Mapping from label IDs to label names
Returns:
Function compatible with HuggingFace Trainer
"""
def compute_metrics_fn(eval_pred):
return compute_metrics(eval_pred, id2label)
return compute_metrics_fn
def plot_confusion_matrix(
y_true: np.ndarray,
y_pred: np.ndarray,
labels: List[str],
save_path: str = "confusion_matrix.png",
normalize: bool = False,
figsize: Tuple[int, int] = (10, 8)
) -> None:
"""
Plot and save confusion matrix with optional normalization.
Args:
y_true: True labels
y_pred: Predicted labels
labels: List of label names
save_path: Path to save the plot
normalize: If True, normalize confusion matrix to percentages
figsize: Figure size (width, height)
"""
cm = confusion_matrix(y_true, y_pred, labels=list(range(len(labels))))
if normalize:
cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]
fmt = '.2f'
title = 'Normalized Confusion Matrix'
else:
fmt = 'd'
title = 'Confusion Matrix'
plt.figure(figsize=figsize)
sns.heatmap(
cm,
annot=True,
fmt=fmt,
cmap='Blues',
xticklabels=labels,
yticklabels=labels,
cbar_kws={'label': 'Percentage' if normalize else 'Count'}
)
plt.title(title, fontsize=14, fontweight='bold')
plt.ylabel('True Label', fontsize=12)
plt.xlabel('Predicted Label', fontsize=12)
plt.tight_layout()
# Create directory if it doesn't exist
os.makedirs(os.path.dirname(save_path) if os.path.dirname(save_path) else '.', exist_ok=True)
plt.savefig(save_path, dpi=300, bbox_inches='tight')
plt.close()
print(f"Confusion matrix saved to {save_path}")
def print_classification_report(
y_true: np.ndarray,
y_pred: np.ndarray,
labels: List[str],
output_dict: bool = False
) -> Optional[Dict]:
"""
Print detailed classification report.
Args:
y_true: True labels
y_pred: Predicted labels
labels: List of label names
output_dict: If True, return report as dictionary instead of printing
Returns:
Classification report as dictionary if output_dict=True, else None
"""
report = classification_report(
y_true,
y_pred,
target_names=labels,
digits=4,
output_dict=output_dict,
zero_division=0
)
if output_dict:
return report
print("\nClassification Report:")
print("=" * 60)
print(report)
return None
def plot_training_curves(
train_losses: List[float],
eval_losses: List[float],
eval_metrics: Dict[str, List[float]],
save_path: str = "./results/training_curves.png"
) -> None:
"""
Plot training and evaluation curves.
Args:
train_losses: List of training losses per step/epoch
eval_losses: List of evaluation losses per step/epoch
eval_metrics: Dictionary of metric names to lists of values
save_path: Path to save the plot
"""
fig, axes = plt.subplots(2, 2, figsize=(15, 10))
# Loss curves
axes[0, 0].plot(train_losses, label='Train Loss', color='blue')
axes[0, 0].plot(eval_losses, label='Eval Loss', color='red')
axes[0, 0].set_xlabel('Step/Epoch')
axes[0, 0].set_ylabel('Loss')
axes[0, 0].set_title('Training and Validation Loss')
axes[0, 0].legend()
axes[0, 0].grid(True, alpha=0.3)
# Accuracy
if 'accuracy' in eval_metrics:
axes[0, 1].plot(eval_metrics['accuracy'], label='Accuracy', color='green')
axes[0, 1].set_xlabel('Step/Epoch')
axes[0, 1].set_ylabel('Accuracy')
axes[0, 1].set_title('Validation Accuracy')
axes[0, 1].legend()
axes[0, 1].grid(True, alpha=0.3)
# F1 Score
if 'f1_weighted' in eval_metrics:
axes[1, 0].plot(eval_metrics['f1_weighted'], label='F1 (weighted)', color='purple')
axes[1, 0].set_xlabel('Step/Epoch')
axes[1, 0].set_ylabel('F1 Score')
axes[1, 0].set_title('Validation F1 Score')
axes[1, 0].legend()
axes[1, 0].grid(True, alpha=0.3)
# Precision and Recall
if 'precision_weighted' in eval_metrics and 'recall_weighted' in eval_metrics:
axes[1, 1].plot(eval_metrics['precision_weighted'], label='Precision', color='orange')
axes[1, 1].plot(eval_metrics['recall_weighted'], label='Recall', color='cyan')
axes[1, 1].set_xlabel('Step/Epoch')
axes[1, 1].set_ylabel('Score')
axes[1, 1].set_title('Validation Precision and Recall')
axes[1, 1].legend()
axes[1, 1].grid(True, alpha=0.3)
plt.tight_layout()
os.makedirs(os.path.dirname(save_path) if os.path.dirname(save_path) else '.', exist_ok=True)
plt.savefig(save_path, dpi=300, bbox_inches='tight')
plt.close()
print(f"Training curves saved to {save_path}")
|