haotian-liu/LLaVA

Issues

• [Question] Differences between forward and generate methods

  #1819 opened 2 days ago by victorcaquilpan
  0

• [Usage] Failed to start stage 1 training of LLaVA-v1.5-7b

  #1818 opened 5 days ago by whyisverysmart
  0

• [Usage] Error in loading mistral 7b weights

  #1817 opened 6 days ago by bryanlinnan
  0

• 使用finetune_task_lora.sh微调llava_1.5模型无法运行

  #1811 opened 22 days ago by zhangye0402
  1

• [Discussion] Newbie alert - asked below question llava:13b about guitar scales and its giving wrong information.

  #1816 opened 6 days ago by ginamdar
  0

• Can llava-v1.5-7b accept the images smaller than 336? without resize, just direct go into the network?

  #1815 opened 7 days ago by 924973292
  0

• [Question] Fintune llava-v1.5-7b just use all textvqa data, but get Accuracy: 0.00%

  #1789 opened a month ago by ethanyys
  1

• [Question] How do I set the image input to a null value?

  #1814 opened 16 days ago by rickeyhhh
  0

• [Usage] RuntimeError: Failed to import transformers.trainer

  #1808 opened 25 days ago by EEElisa
  2

• [Question] Can not reproduce LLaVA 1.5 performance on ScienceQA

  #1770 opened 2 months ago by yiwei-chenn
  1

• [Usage] Cannot Launch a gradio web server

  #1812 opened 22 days ago by LiuChang-ao
  0

• 使用finetune_task_lora.sh微调llava_1.5模型无法运行

  #1810 opened 22 days ago by zhangye0402
  0

• [Usage] Issue with Sampling and Beam Search in generate for LLaVA Models

  #1809 opened 24 days ago by Liumx2020
  0

• [Usage] Inference Speed Issue with LoRA Fine-tuned Model on ScienceQA

  #1763 opened 2 months ago by jinghanSunn
  1

• [Question] Abnormal Generation after finetuning ONLY single label classification dataset

  #1796 opened a month ago by enkaranfiles
  0

• [Question] How do you fine-tune LLaVA-NeXT on video data?

  #1795 opened a month ago by DrVictorBenjamin
  1

• [Usage] peft 1.40.0 requires transformers at least 0.47

  #1794 opened a month ago by shtu-ryan
  0

• [Question] Model parameters during finetuning (prints me only mm_projector parameters)

  #1784 opened a month ago by daulettoibazar
  1

• [Question] Why does text-only data use the empty image token?

  #1792 opened a month ago by MSungK
  0

• [Usage] Missing file 'model_vqa_qbench'

  #1764 opened 2 months ago by eslambakr
  1

• [Question] grad_norm=None

  #1787 opened a month ago by Pixel-anter
  0

• Issue with 4-bit Quantization for LLaVA-NeXT-Video-32B Model on A100-40GB GPU

  #1791 opened a month ago by Rachel0901
  0

• [Question] When I reproduced the second stage using finetune.sh, the processing time per image was too slow

  #1790 opened a month ago by TanmouTT
  1

• [Usage] finetune_task_lora.sh -> Error "exits with return code = -7"

  #1788 opened a month ago by matsutaku44
  2

• [Usage] Infer on model finetuned using finetune_qlora.sh

  #1771 opened 2 months ago by amagzari
  1

• [Question] demo website error https://llava.hliu.cc/

  #1786 opened a month ago by zkailinzhang
  0

• [Question] llava 预训练中断，继续预训练默认加载模型报错

  #1785 opened a month ago by cqray1990
  0

• [Question] Does llava support dynamic image tokens input?

  #1783 opened a month ago by hktk07
  0

• [Discussion] loss function of finetune llava 1.5 with sft

  #1782 opened a month ago by bollossom
  0

• [Question] where can download OCR-VQA data?

  #1781 opened 2 months ago by cqray1990
  0

• [Question] when run v1.5/pretrain.sh , there are some errors

  #1780 opened 2 months ago by cqray1990
  0

• [Question] pretrain data

  #1779 opened 2 months ago by cqray1990
  0

• [Question] I trained the llava-1.5-13B model, but when evaluating, the inference answer is always empty and the speed is very slow.

  #1778 opened 2 months ago by ykzqjyhhh
  0

• [Usage] Batch evaluation using sqlang doesn't support llava-v1.5 model

  #1777 opened 2 months ago by pspdada
  0

• [Question] Where can I obtain the training dataset of the LLava 1.5?

  #1776 opened 2 months ago by weiaicunzai
  1

• [Usage] json.decoder.JSONDecodeError: Expecting ',' delimiter: line 1559608 column 82 (char 39444904)

  #1775 opened 2 months ago by tianke0711
  0

• [Question] Image-text match

  #1774 opened 2 months ago by hazardout
  0

• [Usage] The issue encountered when fine-tuning llava_mistral1.6 using LoRA

  #1772 opened 2 months ago by yuwang4321
  0

• Where is convert_answer_to_mme.py?

  #1769 opened 2 months ago by Tramac
  0

• [Question] disable print

  #1768 opened 2 months ago by LiXinYuann
  0

• [Question] LLaVA 1.5 7B model fine-tune -- pydantic

  #1765 opened 2 months ago by yiwei-chenn
  1

• [Question] Pretrain preprocess

  #1767 opened 2 months ago by leo-young
  0

• [Question] Issue with trying to reproduce the results of LLaVA-Bench-in-the-Wild on LLaVA-v1.5-7b. Has anyone who's reproduced it got this error?

  #1766 opened 2 months ago by cookiesupers22
  0

• [Usage] Error during evaluation for image.../image.png: Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument mat1 in method wrapper_CUDA_addmm)

  #1762 opened 2 months ago by ChathurangiShyalika
  0

• Fine tune for object detection

  #1761 opened 2 months ago by ck-amrahd
  0

• [Discussion] pip install e . error

  #1760 opened 2 months ago by JavaWebT
  0

• [Usage] Training process get stuck in the last iteration of instruction finetuning phrase.

  #1759 opened 2 months ago by fmy7834
  0

• [Usage] How to run CLI after Visual Instruction Tuning with lora?

  #1757 opened 2 months ago by YUECHE77
  1

• [Question] Is it possible to extract the latent representation of the image input from the model?

  #1758 opened 2 months ago by Tizzzzy
  0

• Could you please provide the test files of vqav2 and gqa on llava-15-7b？

  #1756 opened 2 months ago by liuting20
  0