InternVL
InternVL copied to clipboard
[Feature] 使用--use_packed_ds True 时,会报错,请问该如何修改。
Motivation
File "internvl_chat/internvl/patch/pad_data_collator.py", line 104, in concat_pad_data_collator batch[k] = torch.stack([f[k] for f in features]) RuntimeError: stack expects each tensor to be equal size, but got [8145] at entry 0 and [8153] at entry 1
Related resources
No response
Additional context
No response
请问这个问题解决了吗
请问这个问题解决了吗