Update README.md

README.md

@@ -14,6 +14,14 @@ The SuperSparse-Mixtral Large Language Model (LLM) is an sparsified version of t
 
 Our code for accelerating SuperSparse-Mixtral is currently being refined. Stay tuned! Now you can run this model like dense model.
 
+## Chat-Template
+
+During sparsification, we also utilize some SFT datasets.
+We take ChatML as our chat template:
+```
+<|im_start|>user\n{{content}}<|im_end|>\n<|im_start|>assistant\n
+```
+
 ## Allow Finetuning
 
 As we merged the predictors for FFN neurons in models, you can finetune SuperSparse-Mixtral with any framework and algorithm.