Update README.md

README.md

@@ -3,6 +3,8 @@ base_model: meta-llama/Meta-Llama-3.1-8B-Instruct
 library_name: peft
 datasets:
 - sanaa-11/math-dataset
+language:
+- fr
 ---
 # Model Card for LLaMA 3.1 Fine-Tuned Model
 
@@ -91,7 +93,7 @@ for _ in range(5):
 ## Training Details
 
 ### Training Data
-- **Dataset**: The model was fine-tuned on a custom dataset consisting of 11,106 rows of math exercises, lesson content, and solutions, specifically designed for Moroccan students in French.
+- **Dataset**: The model was fine-tuned on a custom dataset consisting of 3.6K rows of math exercises, lesson content, and solutions, specifically designed for Moroccan students in French laungage.
 
 ### Training Procedure
 
@@ -102,9 +104,9 @@ for _ in range(5):
 ### Training Hyperparameters
 - **Training Regime**: The model was fine-tuned using 4-bit quantization with QLoRA to optimize GPU and RAM usage. The training was performed on a Kaggle environment with limited resources.
 - **Batch Size**: 1 (with gradient accumulation steps of 8)
-- **Number of Epochs**: 10
+- **Number of Epochs**: 8
 - **Learning Rate**: 5e-5
-- **Optimizer**: AdamW
+
 
 ## Evaluation
 
@@ -126,7 +128,7 @@ for _ in range(5):
 ### Summary
 **Model Examination**
 - The model demonstrated a consistent reduction in both training and validation loss across the training epochs, suggesting effective learning and generalization from the provided dataset.
-- While F1 score and perplexity were not used in this evaluation, the training and validation losses provide a strong indication of the model's performance and its potential for generating accurate and relevant math exercises.
+
 
 ## Environmental Impact
 **Carbon Emissions**