Update README.md

README.md

@@ -2,6 +2,9 @@
 base_model: meta-llama/Llama-3.2-1B-Instruct
 datasets:
 - JunxiongWang/sftdatasetv3
+- HuggingFaceH4/ultrafeedback_binarized
+- HuggingFaceH4/orca_dpo_pairs
+- JunxiongWang/llama3-ultrafeedback-armorm
 model-index:
 - name: X-EcoMLA-1B1B-dynamic-0.95-DPO 
   results: []
@@ -34,6 +37,16 @@ The X-EcoMLA models are not trained from scratch. Instead, they are composed fro
 | 3. SFT            | End-to-End Knowledge Distillation     | The initialized model is fine-tuned via knowledge distillation.                                                  |  
 | 4. Alignment      | Direct Preference Optimization (DPO)  | In the final stage, DPO is used to align the model's preferences, with the distilled student model itself serving as the reference model for stability.                                      | 
 
+## Training Data 
+
+|Stage      | Dataset                                                                   | License                |   
+|-----------|---------------------------------------------------------------------------|------------------------|
+| SFT       | https://huggingface.co/datasets/teknium/OpenHermes-2.5                    | Refer source materials | 
+| SFT       | https://huggingface.co/datasets/tomg-group-umd/GenQA                      | CC BY-NC 4.0           | 
+| SFT       | https://huggingface.co/datasets/BAAI/Infinity-Instruct                    | CC BY-SA 4.0           | 
+| DPO       | https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized     | MIT                    | 
+| DPO       | https://huggingface.co/datasets/HuggingFaceH4/orca_dpo_pairs              | MIT                    |  
+| DPO       | https://huggingface.co/datasets/JunxiongWang/llama3-ultrafeedback-armorm  | MIT                    | 
 
 ## Getting Started