WFJ

 Thank you very much for sharing your works! Is it convenient to release the relevant code for the experimental results of img2llm? This folder appears empty.Looking forward to your...

Data download

3

 Follow the link of the ocr data set to download the relevant image, but when I execute the fine-tuning instruction CUDA_VISIBLE_DEVICES=0,1 bash scripts/tiny_llava/finetune/finetune_lora.sh, I always encounter the above problems....

loss does not decrease

6

Hello, thank you very much for sharing your work. In the TinyLLaVA_Factory-main file, I executed bash ./scripts/train/train_phi.sh and found a problem. Its loss has been around 5 and has not...

Qwen2-7B for pre-training and fine-tuning

1

Thanks for sharing your work, it's great. Can we use Qwen2-7B for pre-training and fine-tuning? Or Phi-3 or Phi-4?