Update app.py

app.py

@@ -201,11 +201,8 @@ def run(message: dict, history: list[dict], system_prompt: str = "", max_new_tok
     inputs = {k: v.to(device=model.device, dtype=torch.bfloat16) for k, v in raw_inputs.items()}
 
     streamer = TextIteratorStreamer(processor, timeout=30.0, skip_prompt=True, skip_special_tokens=True)
-    generate_kwargs = dict(
-        inputs=inputs,
-        streamer=streamer,
-        max_new_tokens=max_new_tokens,
-    )
+    # Unpack inputs into generate_kwargs so that each tensor is passed as a separate keyword argument.
+    generate_kwargs = {**inputs, "streamer": streamer, "max_new_tokens": max_new_tokens}
     # Launch generation in a separate thread.
     t = Thread(target=generate_thread, kwargs={"generate_kwargs": generate_kwargs})
     t.start()