Add model card with evaluation matrix, confusion matrix, and per-class metrics

README.md

@@ -0,0 +1,48 @@
+
+# Fine-Tuned LLaMA-3-8B CEFR Model
+
+This is a fine-tuned version of `unsloth/llama-3-8b-instruct-bnb-4bit` for CEFR-level sentence generation.
+
+- **Base Model**: unsloth/llama-3-8b-instruct-bnb-4bit
+- **Fine-Tuning**: LoRA with SMOTE-balanced dataset
+- **Training Details**:
+  - Dataset: CEFR-level sentences with SMOTE and undersampling for balance
+  - LoRA Parameters: r=32, lora_alpha=32, lora_dropout=0.5
+  - 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**:
+  - CEFR Classifier Accuracy: 0.250
+  - Precision (Macro): 0.130
+  - Recall (Macro): 0.250
+  - F1-Score (Macro): 0.153
+  - Perplexity: 14.218
+  - Diversity (Unique Sentences): 0.933
+  - Inference Time (ms): 2242.946
+  - Model Size (GB): 4.8
+  - Robustness (F1): 0.145
+- **Confusion Matrix**:
+  - CSV: [confusion_matrix.csv](confusion_matrix.csv)
+  - Image: [confusion_matrix.png](confusion_matrix.png)
+- **Per-Class Confusion Metrics**:
+  - 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
+  - B2: TP=2, FP=7, FN=8, TN=43
+  - C1: TP=3, FP=7, FN=7, TN=43
+  - C2: TP=0, FP=0, FN=10, TN=50
+- **Usage**:
+  ```python
+  from transformers import AutoModelForCausalLM, AutoTokenizer
+
+  model = AutoModelForCausalLM.from_pretrained("Mr-FineTuner/Test___01_withNewEval")
+  tokenizer = AutoTokenizer.from_pretrained("Mr-FineTuner/Test___01_withNewEval")
+
+  # Example inference
+  prompt = "<|user|>Generate a CEFR B1 level sentence.<|end|>"
+  inputs = tokenizer(prompt, return_tensors="pt")
+  outputs = model.generate(**inputs, max_length=50)
+  print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+  ```
+
+Uploaded using `huggingface_hub`.