LoreonLabsPlatform

Overview

Intelligence

Markets
Builders
Research
Ecosystems
Launchpads

Other

Qwen2-VL-Finetune

An open-source implementaion for fine-tuning Qwen2-VL and Qwen2.5-VL series by Alibaba Cloud.

OtherEmerging

Stars

—

Forks

—

Contributors

1

Last push

16mo ago

Recent commits

Latest commits.

fix: Upgrade the version of huggingface-hub to resolove conflict with latest transformers version.
95296152U116mo ago
docs: Update README for proeprly guide to install the transformers version.
ac65ee22U116mo ago
feat: Add option to use liger kernel for properly supporting QLoRA.
8d3fb7a2U116mo ago
fix: Resolve deadlock when full-finetuning with mixed-modality data.
567a6252U116mo ago
style: fixed for creating dummies only when `input_embeds` are `None`.
50fcbd02U116mo ago

style: Add comment about dummy_pixel and dummy_grid.

087263c2U116mo ago

fix: properly add dummy to cuda graph when one shard has video+image in the batch and other dosen't.

1d62f842U116mo ago

docs: Add description for using adamw_8bit_optimizer.

f6137fd2U116mo ago

Top contributors

Builders behind this project.