Add model card with evaluation matrix, confusion matrix, and within-1 metrics

README.md

@@ -11,20 +11,26 @@ This is a fine-tuned version of `unsloth/llama-3-8b-instruct-bnb-4bit` for CEFR-
   - Training Args: learning_rate=2e-5, batch_size=8, epochs=0.1, cosine scheduler
   - Optimizer: adamw_8bit
   - Early Stopping: Patience=3, threshold=0.01
-- **Evaluation Metrics**:
+- **Evaluation Metrics (Exact Matches)**:
   - CEFR Classifier Accuracy: 0.250
   - Precision (Macro): 0.130
   - Recall (Macro): 0.250
   - F1-Score (Macro): 0.153
+- **Evaluation Metrics (Within ±1 Level)**:
+  - CEFR Classifier Accuracy: 0.733
+  - Precision (Macro): 0.701
+  - Recall (Macro): 0.733
+  - F1-Score (Macro): 0.687
+- **Other Metrics**:
   - Perplexity: 14.218
   - Diversity (Unique Sentences): 0.933
-  - Inference Time (ms): 2242.946
+  - Inference Time (ms): 2208.386
   - Model Size (GB): 4.8
   - Robustness (F1): 0.145
-- **Confusion Matrix**:
+- **Confusion Matrix (Exact Matches)**:
   - CSV: [confusion_matrix.csv](confusion_matrix.csv)
   - Image: [confusion_matrix.png](confusion_matrix.png)
-- **Per-Class Confusion Metrics**:
+- **Per-Class Confusion Metrics (Exact Matches)**:
   - A1: TP=0, FP=2, FN=10, TN=48
   - A2: TP=0, FP=0, FN=10, TN=50
   - B1: TP=10, FP=29, FN=0, TN=21