Star152
238
代码介绍
代码
Issues74
Pull Requests29
流水线
Actions
讨论
Wiki
项目成员42
分析
项目设置
Star152
238
  1. MindSpeed-MM
  2. /
  3. mindspeed_mm
  4. /
  5. models
  6. /
  7. transformers
  8. /
  9. qwen3omni
ascend-robotascend-robotfeat(torch): Qwen3-Omni support ulysses cp / fix(torch): repeat_kv and activation_offload bug
e2bc6246创建于 3月2日历史提交
文件最后提交记录最后更新时间
__init__.py
[Modify] Qwen3Omni: optimize memory with FlashAttention-2, activation offload and dynamic chunk loss4 个月前
modeling_outputs.py
[Feature] Initialize Qwen3-Omni model code from transformers library3 个月前
modeling_qwen3_omni_moe.py
feat(torch): Qwen3-Omni support ulysses cp / fix(torch): repeat_kv and activation_offload bug3 个月前
modules.py
feat(torch): Qwen3-Omni support ulysses cp / fix(torch): repeat_kv and activation_offload bug3 个月前
qwen3omni.py
[Refactor]Refactor qwen3omni attention and use fusion op3 个月前