Issues

• Verify evals on Papers with Code

  haotian-liu/LLaVA#1932 · Filter by author NielsRogge opened on Sep 14, 2026

• [Question] Why does LLaVA evaluation assert batch_size == 1 for benchmark inference?

  haotian-liu/LLaVA#1928 · Filter by author AustinKol opened on May 24, 2026

• [Question] What are the dependencies like what is the minimal Flashattention verison, Pytorch version any other ?

  haotian-liu/LLaVA#1927 · Filter by author rangapv opened on May 22, 2026

• [Question] the number of finetuning epochs mismatch

  haotian-liu/LLaVA#1926 · Filter by author Yusen-Peng opened on Apr 18, 2026

• LLaVA-v1.5-7B fails to parse images and returns no valid response in Gradio UI on Windows+RTX 5060

  haotian-liu/LLaVA#1922 · Filter by author wutongshu-6 opened on Feb 27, 2026

• Help with "assert self.weight.shape[1] == 1" Error in LLaVA After Fine-Tuning

  haotian-liu/LLaVA#1919 · Filter by author bellenunesbr opened on Jan 21, 2026

• Tokenizer behavior difference when adding custom token "[SEG]" between LLaVA v1.5 and LLaVA-LLaMA2

  haotian-liu/LLaVA#1918 · Filter by author wannnian opened on Jan 20, 2026

• [Usage] EvalAI challenge server for vizwiz is no longer open for testing

  haotian-liu/LLaVA#1916 · Filter by author ShristiDasBiswas opened on Dec 19, 2025
  3

• [Usage] Demo link

  haotian-liu/LLaVA#1915 · Filter by author burcia1711 opened on Nov 14, 2025

• [Discussion] How to finetune_task_lora.sh in google collab pro which is already python 3.12?

  haotian-liu/LLaVA#1914 · Filter by author anatasiasb30 opened on Nov 5, 2025
  3

• [Question] Changing the vision tower.

  haotian-liu/LLaVA#1913 · Filter by author ronaldAlvxs opened on Oct 29, 2025
  1

• [Usage] Is the input image required to be a multiple of 32?

  haotian-liu/LLaVA#1907 · Filter by author ZhijiangTang opened on Sep 18, 2025

• [Question] Token indices sequence length is longer than the specified maximum sequence length for this model (2071 > 2048).

  haotian-liu/LLaVA#1906 · Filter by author liuyafeng0609 opened on Sep 17, 2025
  1

• [Question] Cannot reproduce LLaVA-v1.5-7b performance on MM-VET

  haotian-liu/LLaVA#1905 · Filter by author Wuzimeng opened on Sep 15, 2025

• [Question] How to use Qwen3 as the base LLM in LLaVA?

  haotian-liu/LLaVA#1903 · Filter by author jane70789-cyber opened on Aug 26, 2025

• [Usage] Model sometimes refuses to describe certain images even with objective, non-contextual prompts

  haotian-liu/LLaVA#1900 · Filter by author urlan opened on Aug 15, 2025
  1

• [Question] Got bad performance after pretraining and finetuning LLaVA-1.5-7B with clip-vit-large-patch14 (224px resolution)

  haotian-liu/LLaVA#1899 · Filter by author ThisisBillhe opened on Jul 17, 2025

• [Question] Why set use_cache=false leads to "RuntimeError: The size of tensor a (638) must match the size of tensor b (639) at non-singleton dimension 2"

  haotian-liu/LLaVA#1898 · Filter by author 924973292 opened on Jul 11, 2025

• [Usage] How to resolve the conflict between the requirement to use CUDA 12.8 with PyTorch 2.7.0 for 5090 and the environment conflict caused by the -e. flag during LLaVA installation?

  haotian-liu/LLaVA#1897 · Filter by author Lucas-Jin-Qh opened on Jul 10, 2025

• [Usage] May I ask if there is a reasoning script for visual question answering? The script I generated with the large model keeps reporting errors

  haotian-liu/LLaVA#1896 · Filter by author leonh666 opened on Jul 6, 2025

• [Discussion] Llava 1.5 model won't generate caption

  haotian-liu/LLaVA#1895 · Filter by author timoty-Joel opened on Jul 4, 2025
  1

• [Usage] Could not parse server response: SyntaxError:JSON.parse: unexpected character at line 1column 1 ofthe JSON data

  haotian-liu/LLaVA#1894 · Filter by author ZhangJinian opened on Jun 27, 2025
  2

• [Usage] UserWarning: c10d::broadcast_: an autograd kernel was not registered to the Autograd key(s) but we are trying to backprop through it.

  haotian-liu/LLaVA#1893 · Filter by author hjw1 opened on Jun 24, 2025

• "AttributeError: 'NoneType' object has no attribute 'shape' in llava_arch.py prepare_inputs_labels_for_multimodal" after fine tuning

  haotian-liu/LLaVA#1889 · Filter by author samundiswary-cloud opened on Jun 14, 2025
  3

• [Question] how to lora this llava-hf/llava-1.5-7b-hf model

  haotian-liu/LLaVA#1887 · Filter by author Amark-cheey opened on Jun 7, 2025
  1