Chenyu Wu
Chenyu Wu
I wrote image generation finetuning code. https://github.com/ladvu/Janus I am not sure if there were any bugs. It is just my coursework. So just for your reference...
One possible reason is that ```python sample["seq_len"] % sr * chunk_duration == 0 ``` which results in ```python seq_len_list = [chunk_duration*fps] * (len(audio_chunks) - 1) + [sample['seq_len'] % (chunk_duration*fps)] #...