OpenNLPLab commited on
Commit
cf7a5f5
1 Parent(s): 9cb6819

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +1 -0
README.md CHANGED
@@ -109,6 +109,7 @@ export use_triton=False
109
  >>> pred = model.generate(**inputs, max_new_tokens=2048, repetition_penalty=1.0)
110
  >>> print(tokenizer.decode(pred.cpu()[0], skip_special_tokens=True))
111
  ```
 
112
 
113
 
114
  # Fine-tuning the Model
 
109
  >>> pred = model.generate(**inputs, max_new_tokens=2048, repetition_penalty=1.0)
110
  >>> print(tokenizer.decode(pred.cpu()[0], skip_special_tokens=True))
111
  ```
112
+ * **Note**: we recommend to use `bfloat16` in `TransNormerLLM`, `float16` might lead `nan` error, please check your divce compatibility!
113
 
114
 
115
  # Fine-tuning the Model