Update fine_tune_inference_test_mistral.py

fine_tune_inference_test_mistral.py

@@ -76,8 +76,7 @@ def chat(msg: Message):
             return {"error": "Boş giriş"}
 
         messages = [{"role": "user", "content": user_input}]
-        inputs = tokenizer.apply_chat_template(messages, return_tensors="pt", add_generation_prompt=True)
-        inputs = inputs.to(model.device)
+        input_ids = tokenizer.apply_chat_template(messages, return_tensors="pt", add_generation_prompt=True).to(model.device)
 
         generate_args = {
             "max_new_tokens": 128,
@@ -94,7 +93,7 @@ def chat(msg: Message):
             })
 
         with torch.no_grad():
-            output = model.generate(**inputs, **generate_args)
+            output = model.generate(input_ids=input_ids, **generate_args)
 
         decoded = tokenizer.decode(output.sequences[0], skip_special_tokens=True)
         answer = decoded.split("</s>")[-1].strip()