change model card

README.md

@@ -15,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # zephyr-7b-sft-lora
 
-This model is a fine-tuned version of [mrs-7b](https://huggingface.co/mrs-7b) on the gpt4_reward_with_format dataset.
+This model is a fine-tuned version of zephyr-7b on the gpt4_reward.
 It achieves the following results on the evaluation set:
 - Loss: 0.0911