Fix generate: apply_chat_template return_dict + unpack (attention mask)

app.py

@@ -59,13 +59,13 @@ def _messages(message, history, system_prompt):
 @spaces.GPU(duration=120)
 def chat(message, history, system_prompt, temperature, max_new_tokens):
     msgs = _messages(message, history, system_prompt)
-    input_ids = tokenizer.apply_chat_template(
-        msgs, add_generation_prompt=True, return_tensors="pt"
+    inputs = tokenizer.apply_chat_template(
+        msgs, add_generation_prompt=True, return_tensors="pt", return_dict=True
     ).to(model.device)
 
     streamer = TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
     gen_kwargs = dict(
-        input_ids=input_ids,
+        **inputs,
         streamer=streamer,
         max_new_tokens=int(max_new_tokens),
         do_sample=temperature > 0,