make gemma's output better (#10943)

2024-05-08 14:27:51 +08:00 · 2024-05-08 14:27:51 +08:00 · 5973d6c753
commit 5973d6c753
parent 15ee3fd542
1 changed files with 1 additions and 0 deletions
--- a/python/llm/example/GPU/HF-Transformers-AutoModels/Model/gemma/generate.py
+++ b/python/llm/example/GPU/HF-Transformers-AutoModels/Model/gemma/generate.py
@ -48,6 +48,7 @@ if __name__ == '__main__':
                                                 load_in_4bit=True,
                                                 optimize_model=True,
                                                 trust_remote_code=True,
+                                                 mixed_precision=True,
                                                 use_cache=True)
    model = model.to('xpu')