Spaces:

bstraehle
/

sft

Running

bstraehle commited on Jul 13

Commit

825f16a

•

1 Parent(s): e2e0492

Update app.py

Files changed (1) hide show

app.py CHANGED Viewed

@@ -56,7 +56,7 @@ def train_model(model_id):
     model = AutoModelForCausalLM.from_pretrained(
         model_id,
         device_map="auto",
-        attn_implementation="flash_attention_2",
         torch_dtype=torch.bfloat16,
         quantization_config=bnb_config
     )

     model = AutoModelForCausalLM.from_pretrained(
         model_id,
         device_map="auto",
+        #attn_implementation="flash_attention_2",
         torch_dtype=torch.bfloat16,
         quantization_config=bnb_config
     )