coder4nlp

[Question] When will LLaVA-NeXT be open sourced?

3

### Question _No response_

DocOwl1.5 training code?

12

When will the training code be released?Thx.

The interleaved data from OmniCorpus was also utilized. Could you please provide an example to illustrate the data processing process? Thanks!

The inference time of qwen2.5-vl is very slow.

36

Dashinfer is 10 times slower than VLLM. How can this issue be resolved?

Distill

1

请问要准备的蒸馏数据，num_inference_steps,cfg_scale表示要蒸馏到对应的步数和cfg吗，还是说需要使用diffusiondb数据集中的原始数据50 | 7

Because I am unable to register on Discord, as it is always verifying the human identity.