Update README.md

README.md

@@ -22,12 +22,13 @@ model-index:
     metrics:
       - name: Loss
         type: loss
-        value: 4.4
+        value: 4.69
 ---
 
 # T5-Small with LoRA on OpenCodeReasoning
 
 This is a LoRA fine-tuned version of T5-small on a subset of NVIDIA's OpenCodeReasoning dataset using [PEFT](https://github.com/huggingface/peft).
+Improved version to be uploaded soon. 
 
 ## Loss Curve
 
@@ -43,7 +44,8 @@ This is a LoRA fine-tuned version of T5-small on a subset of NVIDIA's OpenCodeRe
 | 400  | 4.89       | 4.42     |
 | 450  | 4.69       | 4.40     |
 
-Final Train Loss: **5.71**
+Final Train Loss: **4.69**
+Final Eval Loss: **4.40**
 
 ## Example Usage
 
@@ -59,8 +61,9 @@ tokenizer = AutoTokenizer.from_pretrained("ShahzebKhoso/t5-small-opencode-lora")
 inputs = tokenizer("generate code: write a function to reverse a string", return_tensors="pt")
 outputs = model.generate(**inputs)
 print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+'''
 
-Notes
+## Notes
 
 Trained on subset of OpenCodeReasoning due to Colab memory limits
 
@@ -69,6 +72,6 @@ Use PeftModel with t5-small base
 Metrics used: Loss (BLEU skipped due to output structure)
 
 
-License
+## License
 
 Apache 2.0