5 Pull requests merged by 5 people

• Add SD3 fine-tuning scripts

  #1966 merged Jun 3, 2025

• Update readme files for explicit lazy mode

  #1921 merged Jun 3, 2025

• Enable mixtral 8x7b accuracy evaluation

  #1986 merged Jun 3, 2025

• Hot fix regional compilation

  #2005 merged Jun 2, 2025

• [Llama-Vision] Trim logits

  #2009 merged May 29, 2025

5 Pull requests opened by 3 people

• Enable FusedRMSNorm for FLUX

  #2011 opened May 30, 2025

• fea(): updated the readme for mediapipe support

  #2012 opened May 30, 2025

• Fix SD3 flag in README example

  #2013 opened Jun 4, 2025

• Fix deepspeed regional compilation by doing it before engine initialization

  #2014 opened Jun 4, 2025

• Enable distributed CFG for SD3 pipeline

  #2015 opened Jun 5, 2025

8 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.

• Enabling Snowflake Arctic on Gaudi 3

  #1719 commented on Jun 3, 2025 • 0 new comments

• Upgrade to Transformers v4.51

  #1943 commented on May 29, 2025 • 0 new comments

• Add qwen3 family

  #1948 commented on Jun 3, 2025 • 0 new comments

• Llama 3.2 - Fix the issue for eager mode (#260)

  #1976 commented on May 29, 2025 • 0 new comments

• Configure qlora tests with additional arguments

  #1983 commented on Jun 3, 2025 • 0 new comments

• force model preparation on eval if torch.compile enabled

  #1993 commented on Jun 4, 2025 • 0 new comments

• Add groups to slow_tests_image_to_text_example

  #2008 commented on Jun 4, 2025 • 0 new comments

• Use the original TE FP8 defaults

  #2010 commented on May 29, 2025 • 0 new comments