6 Pull requests merged by 4 people

• pixtral fix conversion

  #315 merged Jun 20, 2025

• Fix dropless MOE, add tests for sparse linear,

  #310 merged Jun 19, 2025

• Run tests for all models

  #289 merged Jun 19, 2025

• Testing cleanup: remove redundant tests, reduce memory usage, replace materialize_meta_tensors

  #308 merged Jun 19, 2025

• Fix hybrid export bug

  #313 merged Jun 19, 2025

• Refactoring of Evaluation and adding of evaluate command

  #264 merged Jun 19, 2025

1 Pull request opened by 1 person

• Fix and improve loading of distributed checkpoints

  #314 opened Jun 19, 2025

2 Issues closed by 1 person

• [bug] Multiple test failures when testing all models

  #291 closed Jun 20, 2025

• [bug] MOE is broken with triton 3.2.0, 3.3.0 and 3.3.1

  #303 closed Jun 19, 2025

6 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.

• Fix checkpoint export errors for the Dream model

  #311 commented on Jun 23, 2025 • 5 new comments

• WIP: multimodal support

  #227 commented on Jun 20, 2025 • 2 new comments

• WIP: Multimodal Audio

  #272 commented on Jun 23, 2025 • 0 new comments

• [work in progress] Adds lm_eval to evaluations

  #282 commented on Jun 25, 2025 • 0 new comments

• Masked Diffusion Training

  #294 commented on Jun 25, 2025 • 0 new comments

• Changes for AR + Diffusion training in Fast-LLM

  #312 commented on Jun 23, 2025 • 0 new comments