Update README.md

README.md

@@ -86,14 +86,19 @@ print(bot_message)
 ```
 
 
-# Training datasets
-This model was trained using the ChatML format, so it should be used for inference using the ChatML chatbot format. 
-We chose this format as the base model ([Open-Orca/Mistral-7B-SlimOrca](https://huggingface.co/Open-Orca/Mistral-7B-SlimOrca)) was trained with this format, and we find the chatbot format more compelling for practical use compared to the Alpaca style instruction format.
+# Training details
 
+We trained on the following 3 datasets:
 * [JASTER](https://github.com/llm-jp/llm-jp-eval)
 * [kunishou/oasst1-89k-ja](https://huggingface.co/datasets/kunishou/oasst1-89k-ja/)
 * [kunishou/databricks-dolly-15k-ja](https://huggingface.co/datasets/kunishou/databricks-dolly-15k-ja/)
 
+using the ([Open-Orca/Mistral-7B-SlimOrca](https://huggingface.co/Open-Orca/Mistral-7B-SlimOrca)) model as our base checkpoint.
+
+This model was trained using the ChatML format, so it should be used for inference using the ChatML chatbot format.
+We chose this format as the base model ([Open-Orca/Mistral-7B-SlimOrca](https://huggingface.co/Open-Orca/Mistral-7B-SlimOrca)) was trained with this format, and we find the chatbot format more compelling for practical use compared to the Alpaca style instruction format.
+
+
 We trained for 1 epoch using the following Axolotl config. (Early stopping was not performed during our training.)
 <details><summary>Axolotl config .yaml</summary>