Update app.py

app.py

@@ -110,7 +110,7 @@ async def predict(request: DownloadModelRequest):
             with open(model_path, 'wb') as f:
                 f.write(model_stream.read())
             
-            model = AutoModelForCausalLM.from_pretrained(tmp_dir, from_tf=False)
+            model = AutoModelForCausalLM.from_pretrained(tmp_dir, from_tf=True)
             tokenizer = AutoTokenizer.from_pretrained(tmp_dir)
             
             pipeline_ = pipeline(request.pipeline_task, model=model, tokenizer=tokenizer)