yuzeng
yuzeng
There seem to be some errors in the Dependencies and Installation section.
Thanks for your reply, but still error occurs when configure the environment. And some of the code packages import incorrectly in **projects/DDETRS/train_net.py** e.g. **from detectron2.projects.ddetrs.data.custom_dataset_dataloader import build_custom_train_loader**
I also found it!
File "/cache/zy/VL-RLHF/src/vlrlhf/models/InternLMXC2/__init__.py", line 87, in _merge_input_ids_with_image_features raise ValueError( ValueError: The input provided to the model are wrong. The number of image tokens is 2 while the number of image given...
> File "/cache/zy/VL-RLHF/src/vlrlhf/models/InternLMXC2/**init**.py", line 87, in _merge_input_ids_with_image_features raise ValueError( ValueError: The input provided to the model are wrong. The number of image tokens is 2 while the number of image...
> 我们统一用``作为image token,请检查下输入文本中image token的总数量与输入图片的总数量是否一致这个是没问题的。 我猜测是padding的问题,internXC2的pad_token_id是2,对应embedding为零向量,在执行final_embedding[image_to_overwrite] = image_features.contiguous().reshape(-1, embed_dim).to(target_device)的时候报错
方便展示一下internXC2模型的config.py文件吗
internXC2 的DPO代码还是跑不起来,您确定repo里是最新版本的代码吗?是不是repo是之前的版本导致有些bug没有修复