Update README.md

README.md

@@ -83,6 +83,33 @@ For more experimental results and analysis, please refer to our [report](./HY_MT
 ---
 
 ## Inference and Deployment
+
+For 1.8B and 7B, we recommend using the following parameters for inference. Note that our models do not have a default system_prompt.
+
+```json
+
+{
+  "temperature": 0.7,
+  "top_p": 0.6,
+  "top_k": 20,
+  "repetition_penalty": 1.05,
+  "max_tokens": 4096
+}
+```
+
+For 30B-A3B, we recommend using the following parameters for inference. Note that our models do not have a default system_prompt.
+
+```json
+
+{
+  "temperature": 0.7,
+  "top_p": 1.0,
+  "top_k": -1,
+  "repetition_penalty": 1.0,
+  "max_tokens": 4096
+}
+```
+
 ### transformers
 
 transformers>=5.6.0
@@ -194,20 +221,6 @@ cmake --build build --config Release
 ```
 
 
-For 1.8B and 7B, we recommend using the following parameters for inference. Note that our models do not have a default system_prompt.
-
-```json
-
-{
-  "temperature": 0.7,
-  "top_p": 0.6,
-  "top_k": 20,
-  "repetition_penalty": 1.05,
-  "max_tokens": 4096
-}
-```
-
-
 ## Model Training
 Hy-MT2 provides a complete model training pipeline, supporting both full-parameter fine-tuning and LoRA fine-tuning, as well as multiple DeepSpeed ZeRO configurations and LLaMA-Factory integration.