Spaces:

izumi-lab
/

llama-13b-japanese-lora-v0-1ep

Paused

masanorihirano commited on May 30, 2023

Commit

da3b30c

•

1 Parent(s): 250b795

fix typo

Files changed (1) hide show

app.py CHANGED Viewed

@@ -42,7 +42,7 @@ def load_lora_model(
     model = LlamaForCausalLM.from_pretrained(
         model_path,
         load_in_8bit=load_8bit,
-        device_map="auto " if device == "cuda" else {"": device},
         max_memory=max_gpu_memory,
         torch_dtype=torch.float16,
     )

     model = LlamaForCausalLM.from_pretrained(
         model_path,
         load_in_8bit=load_8bit,
+        device_map="auto" if device == "cuda" else {"": device},
         max_memory=max_gpu_memory,
         torch_dtype=torch.float16,
     )