Search results

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

Status: Open.

#1914 In haotian-liu/LLaVA;

· anatasiasb30 opened on Nov 5, 2025

[Question] Changing the vision tower.

Status: Open.

#1913 In haotian-liu/LLaVA;

· sovietes opened on Oct 29, 2025

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

Status: Open.

#1907 In haotian-liu/LLaVA;

· ZhijiangTang opened on Sep 18, 2025

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

Status: Open.

#1906 In haotian-liu/LLaVA;

· liuyafeng0609 opened on Sep 17, 2025

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

Status: Open.

#1905 In haotian-liu/LLaVA;

· Wuzimeng opened on Sep 15, 2025

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

Status: Open.

#1903 In haotian-liu/LLaVA;

· jane70789-cyber opened on Aug 26, 2025

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

Status: Open.

#1900 In haotian-liu/LLaVA;

· urlan opened on Aug 15, 2025

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

Status: Open.

#1899 In haotian-liu/LLaVA;

· 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"

Status: Open.

#1898 In haotian-liu/LLaVA;

· 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 <code>-e.</code> flag during LLaVA installation?

Status: Open.

#1897 In haotian-liu/LLaVA;

· 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

Status: Open.

#1896 In haotian-liu/LLaVA;

· leonh666 opened on Jul 6, 2025

[Discussion] Llava 1.5 model won't generate caption

Status: Open.

#1895 In haotian-liu/LLaVA;

· timoty-Joel opened on Jul 4, 2025