可以启动，但是输入prompts就会这样。

Please input prompts:
text|image|A bustling medieval market scene with vendors selling exotic goods under colorful tents
Traceback (most recent call last):
File "/home/marsdy/AnyGPT/anygpt/src/infer/cli_infer_base_model.py", line 317, in interact
self.forward(prompt)
File "/home/marsdy/AnyGPT/anygpt/src/infer/cli_infer_base_model.py", line 308, in forward
response = self.response(modality, to_modality, input_data, voice_prompt)
File "/home/marsdy/AnyGPT/anygpt/src/infer/cli_infer_base_model.py", line 264, in response
input_ids = input_ids.to(self.device)
RuntimeError: CUDA error: out of memory
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

Start by reproducing the prompt in src/infer/cli_infer_base_model.py and read the response path around lines 264, 308, and 317. Inspect how inference allocates GPU memory and use the reported CUDA diagnostics if needed. Done should identify the cause of the out-of-memory failure and establish a tested fix or clearly documented resource requirement.