Skip to content

Pull requests: NVIDIA/Megatron-LM

New pull request New
153 Open 239 Closed
153 Open 239 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[Update] Print training log in rank0
#1296 opened Nov 21, 2024 by shijungg Loading…
support qwen2 hf<->mcore ckpt converter
#1290 opened Nov 19, 2024 by wenyujin333 Loading…
Fix: misnamed sharded instead of common in checkpoint
#1289 opened Nov 16, 2024 by prrathi Loading…
Fix: Resolve multimodal model errors and update README usage instructions
#1286 opened Nov 13, 2024 by singleheart Loading…
Set torch.multiprocessing start method as 'spawn'
#1285 opened Nov 12, 2024 by hxdtest Loading…
Fix a bug in optimizer's mix_lr/max_lr when args.override_opt_param_scheduler==True
#1284 opened Nov 12, 2024 by lyuwen Loading…
Huvu/update t5 attentionmasktype
#1273 opened Nov 4, 2024 by huvunvidia Loading…
Update t5_model.py
#1271 opened Nov 2, 2024 by huvunvidia Loading…
Enable huggingface tokenizer
#1268 opened Oct 30, 2024 by msiddaiah Loading…
fix: remove unnecessary trailing comma in statement
#1265 opened Oct 29, 2024 by singleheart Loading…
Enabling LR scaling for a specific layer (ex. down-projection...) during pretraining
#1262 opened Oct 28, 2024 by dhia680 Loading…
3
[ENHANCEMENT] Add support for Apex RMSNorm for use in qk-norm
#1261 opened Oct 28, 2024 by wdevazelhes Loading…
5
Add support to process gzip files
#1260 opened Oct 28, 2024 by puneeshkhanna Loading…
Make it an option to use TransformerEngine activation function in FFN block
#1233 opened Oct 21, 2024 by guyueh1 Loading…
4
[Wrong spelling] Update training.py
#1229 opened Oct 21, 2024 by zyqhnu Loading…
1
Typo fix in readme
#1223 opened Oct 17, 2024 by alexchen4ai Loading…
2
support qwen2 and siglip weight conversion script to enable training …
#1221 opened Oct 16, 2024 by tao-githup Loading…
2
readme spelling correction
#1216 opened Oct 13, 2024 by jonassteinberg1 Loading…
2
[Functions] Support Packed_seq_params in Megatron-LM
#1215 opened Oct 12, 2024 by Baibaifan Loading…
Embedding
#1209 opened Oct 10, 2024 by rachitgarg91 Loading…
Dev/optimizer offloading
#1205 opened Oct 10, 2024 by lostkevin Loading…
fix bugs for multi_latent_attention
#1203 opened Oct 9, 2024 by xqiangx1991 Loading…
Use consistent assert message
#1195 opened Oct 3, 2024 by youzagou Loading…
2
Expose cp_comm_type in ModelParallelConfig
#1160 opened Sep 27, 2024 by zochaoq Loading…
Enabling UCC backend for PP communication
#1157 opened Sep 24, 2024 by youngeunkwon0405 Loading…
ProTip! Add no:assignee to see everything that’s not assigned.