Hoantrbl

> Hi, this shouldn't affect normal generation. If the model repeats or , try adding: > > ``` > "eos_token_id": [ > 151645, >   151643 >  ] > ``` >...

The main develop difficulties are that I can't find the code for transferring the embedding into the clip embedding like: ```python self.text_projection = nn.Parameter(torch.empty(transformer_width, clip_embed_dim)) ``` https://github.com/OpenGVLab/InternVL/blob/main/clip_benchmark/clip_benchmark/models/internvl_c_pytorch/internvl_c.py#L348 which can not...

I decide to retrain the clip projection layer in the `internvl-chat` like `clip_project` or `text_projection` from `internvl-g`. However, when I perform the contrastive learning on the image embedding and want...

> I decide to retrain the clip projection layer like `clip_project` or `text_projection` from `internvl-g`. However, when I perform the contrastive learning on the image embedding and want to gather...

I also face this problem, any replace plans? Can you help me? > > 有什么进展吗？我也遇到了同样的问题 > > Can't solve it, I switched to Swift

> I also face this problem, any replace plans? Can you help me? > > > > 有什么进展吗？我也遇到了同样的问题 > > > > > > Can't solve it, I switched to...

> facing the same issue @Hoantrbl Could you tell me your torch, cuda and flash-attn versions? Thanks torch 2.1.0+cu121 torchaudio 2.1.0+cu121 torchvision 0.16.0+cu121 flash_attn 2.6.3