Upload app.py with huggingface_hub

app.py

@@ -194,19 +194,21 @@ def generate_spec(
         {"role": "user", "content": user_content},
     ]
 
-    inputs = tokenizer.apply_chat_template(
-        messages, tokenize=True, add_generation_prompt=True, return_tensors="pt"
-    ).to(model.device)
+    input_text = tokenizer.apply_chat_template(
+        messages, tokenize=False, add_generation_prompt=True
+    )
+    inputs = tokenizer(input_text, return_tensors="pt").to(model.device)
+    input_len = inputs["input_ids"].shape[-1]
 
     outputs = model.generate(
-        input_ids=inputs,
+        **inputs,
         max_new_tokens=max_tokens,
         temperature=temperature,
         top_p=top_p,
         do_sample=True,
     )
 
-    generated = outputs[0][inputs.shape[-1] :]
+    generated = outputs[0][input_len:]
     return tokenizer.decode(generated, skip_special_tokens=True)