-
Hi, I've read your code and find that you just use the visual encoder and a prediction head during training, and during inference you just upload all the fine-tuned visual encoder and merge them, it s…
-
This is a great job, but now the link doesn't seem to work properly, Could you update the link? I'd like to try it out.
[an example app](https://github.com/UbiquitousLearning/mllm/android)
-
### Feature request
How to make SiglipVisionModel can support Auto map to map to multiple GPUs.
### Motivation
Currently, using MLLM with siglip, the whole model might need automap, since the visio…
-
When I try to inference [internlm-xcomposer2d5-7b](https://huggingface.co/internlm/internlm-xcomposer2d5-7b), I find the bug of size mismatch.
bug details:
```
Traceback (most recent call last):
…
-
Hi,
We have recently released our latest work, `RLAIF-V: Aligning MLLMs through Open-Source AI Feedback for Super GPT-4V Trustworthiness`, on [GitHub](https://github.com/RLHF-V/RLAIF-V) and [arXiv…
-
### Checklist
- [X] 1. I have searched related issues but cannot get the expected help.
- [X] 2. The bug has not been fixed in the latest version.
### Describe the bug
1.session length长度不一致，…
-
Hi, Nice work, and thanks for sharing! May I ask what's the license for the dataset Cambrian-10M and Cambrian-Alignment? Thanks!
-
Add siglip flashattention support?
I noticed that the navit-flashatten-siglip version already added flash attention, would transformers add it support inside ?
-
can I sft MLLM with your train split?
-
Hi there, thanks for the wonderful project!
I have one question related to the checkpoint function and wonder if flex-attn can work with this feature. In `Megatron-LM` I replaced the original attn_…