11 Pull requests merged by 7 people

• fix llava bench . to _

  #1073 merged Feb 25, 2025

• Qwen model

  #1069 merged Feb 25, 2025

• fix open-magvit2 reshape

  #1070 merged Feb 24, 2025

• Support qwen2.5-vl model High performance inference

  #1063 merged Feb 21, 2025

• fix env setting of benchmark

  #1067 merged Feb 21, 2025

• fix qwen2 train when use fake imag

  #1066 merged Feb 21, 2025

• Allow use xformers memory effeicient attention in dy2st for Stable Diffusion

  #1062 merged Feb 20, 2025

• [wip] datacopilot analysis, filter and visualize

  #1060 merged Feb 20, 2025

• fix qwen2.5vl train config

  #1061 merged Feb 20, 2025

• fix qwen2.5vl sdpa infer

  #1059 merged Feb 19, 2025

• [wip] Add ops into datacopilot

  #1055 merged Feb 19, 2025

4 Pull requests opened by 4 people

• [wip] add token log for llava bench

  #1065 opened Feb 19, 2025

• [WIP]update paddlemix doc

  #1068 opened Feb 24, 2025

• commit

  #1071 opened Feb 25, 2025

• add the script about the lora auto paralllel, fixed the ImageCollatorForSeq2Seq input attribute in funtune_qwen2vl_auto.py

  #1072 opened Feb 25, 2025

4 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• PaddleMIX 快乐开源活动 (2025 H1)

  #1046 commented on Feb 20, 2025 • 0 new comments

• FLUX

  #999 commented on Feb 24, 2025 • 0 new comments

• Fix typos

  #1033 commented on Feb 24, 2025 • 0 new comments

• [wip] data loss for ppdocbee

  #1056 commented on Feb 20, 2025 • 0 new comments