Add offload_buffers=True for Qwen-Omni model loading

transcript.py

@@ -100,7 +100,7 @@ def main():
         else:
             # Use Qwen-Omni
             from transformers import Qwen2_5OmniForConditionalGeneration, Qwen2_5OmniProcessor
-            model = Qwen2_5OmniForConditionalGeneration.from_pretrained("Qwen/Qwen2.5-Omni-7B", torch_dtype="auto", device_map="auto", trust_remote_code=True)
+            model = Qwen2_5OmniForConditionalGeneration.from_pretrained("Qwen/Qwen2.5-Omni-7B", torch_dtype="auto", device_map="auto", trust_remote_code=True, offload_buffers=True)
             processor = Qwen2_5OmniProcessor.from_pretrained("Qwen/Qwen2.5-Omni-7B", trust_remote_code=True)
             conversation = [
                 {"role": "system", "content": [{"type": "text", "text": "You are Qwen, a virtual human developed by the Qwen Team, Alibaba Group, capable of perceiving auditory and visual inputs, as well as generating text and speech."}]},