All pull requests

Pull requests

• docs(qwen2_5): fix stale eval conversion command (hf2mcore_qwen2.5_convertor.sh does not exist)Update README.md

  #735·zhaozhaooo19-design opened Aug 25, 2026·

  ·
  1

• Add Best-Fit Decreasing bin packing for pretrain data preprocessing

  #732·amanyara opened Apr 19, 2026·5 tasks·

  ·
  1

• Fix DeepSeek related bugs (#724, #548, #633)

  #731·Nyquist24 opened Mar 14, 2026·

  ·
  3

  1

• Fix qwen_vl key error in dist converter

  #706·ShareLer opened Nov 12, 2025·

  ·

• fix the sequence packing

  #663·Raymondssssss opened Sep 9, 2025·

  ·
  1

• fix error when convert qwen2.5_vl

  #660·chenhaiq opened Sep 3, 2025·

  ·
  1

• Fix MEGATRON_PATCH_PATH in qwen2_5 run_8xH20.sh

  #646·wyjBot opened Aug 11, 2025·

  ·
  1

• Fix: Correct sample idxmap building in SFT data preprocessing script

  #636·Dear-Sloth opened Jul 19, 2025·

  ·
  1

• fix broken path in run_build_idxmap_sft_dataset.sh

  #627·Opdoop opened Jun 26, 2025·

  ·
  1

• update qwen2_5 readme for converting Megatron-Core -> HuggingFace, a…

  #624·changingivan opened Jun 24, 2025Contributor·

  ·

• Fix MEGATRON_PATCH_PATH

  #622·xu-song opened Jun 23, 2025·

  ·

• fix typo

  #582·1195343015 opened May 5, 2025·

  ·
  1

• fix typo for hidden_size

  #580·ShareLer opened Apr 30, 2025·

  ·
  1

• Update hf2mcore_deepseek_v3_moe.py

  #495·lmc8133 opened Mar 7, 2025·

  ·
  1

• Fix typo

  #478·xu-song opened Feb 24, 2025·

  ·
  1

• support moe_layer_freq@qwen_moe

  #440·laohur opened Jan 20, 2025·

  ·
  2

• Mixtral ggemm to hf format

  #230·vlad-karpuhin opened May 15, 2024·

  ·
  2

• mixtral checkpoint converter - grouped gemm for experts MLP support

  #218·vlad-karpuhin opened Apr 29, 2024·

  ·
  2

• Feature/load balance add expert replacement feature for MoE model(mixtral)

  #187·uygnef opened Apr 18, 2024·

  ·
  1