transformers
bdf5fb70 - Skip non-selected experts for qwen3_moe (#38133)

Commit

307 days ago

Skip non-selected experts for qwen3_moe (#38133) * fix(qwen3moe): skip experts with no workload * avoid tolist and also update other moe models * fix: should squeeze 0-dim only

References

#38133 - Skip non-selected experts for qwen3_moe

#59 - Fix attention mask handling in EoMT-DINOv3 converter

#62 - Add initial DEIMv2 model implementation

#65 - Fix RTDetrV2 sine position embedding ordering

#44375 - Add RF-DETR

#71 - Use Mask2Former ignore_value in mask matching and losses

#44385 - Fix make check-repo

#45082 - [VidEoMT] Update conversion script

#45110 - Add SAM 3.1

Author

seven-mile

Parents

719058c6

transformers bdf5fb70 - Skip non-selected experts for qwen3_moe (#38133)

transformers
bdf5fb70 - Skip non-selected experts for qwen3_moe (#38133)