Skip to content

Pull requests: NVIDIA/Megatron-LM

New pull request New
181 Open 282 Closed
181 Open 282 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

CUDA graph fixes for Llama3.1
#1534 opened Apr 14, 2025 by vasunvidia Loading…
[BUGFIX] Save dist_checkpointing metadata on all nodes for multi-node training
#1531 opened Apr 13, 2025 by Pranaykarvi Loading…
1
1
Fix AttributeError in MultiTokenPredictionLayer
#1529 opened Apr 12, 2025 by shenyunhang Loading…
added fix to avoid overflow with new numpy casting behaviour (Issue: #1519)
#1520 opened Apr 4, 2025 by Apsod Loading…
Add full support for Local mode without Apex/TE, and add support for Open XLA on CUDA
#1510 opened Mar 31, 2025 by ajayvohra2005 Loading…
[BUG]: Updating the logic for reducing the load_balancing_loss during logging, such that the correct value is logged while using CUDA Graphs
#1507 opened Mar 27, 2025 by arjun-choudhry Loading…
1
Fix typo on distrib_optimizer.py
#1505 opened Mar 26, 2025 by wplf Loading…
fix for group_limited_topk: K_r is moe_router_topk instead of moe_router_num_groups
#1502 opened Mar 25, 2025 by ladyrick Loading…
fix: MultiLatentAttention cp_comm_type
#1499 opened Mar 24, 2025 by RandMist Loading…
[Bug Fix] fix p2p communication order error and stuck problems when pp 2 and vpp 2 with remove pad
#1495 opened Mar 22, 2025 by ETOgaosion Loading…
1
Fix llama_mistral loader by using args.true_vocab_size
#1491 opened Mar 20, 2025 by zhuzilin Loading…
vscode/cursor devcontainer
#1483 opened Mar 14, 2025 by yzhang123 Loading…
Build dataset for all GPUs with tp_rank=0 and pp_rank=0 or -1 in multi-machine training.
#1480 opened Mar 14, 2025 by wan-nan Loading…
Set hashlib.md5 usedforsecurity=False, #1471
#1472 opened Mar 12, 2025 by jsta Loading…
Enabling variable_seq_lengths when encoder has Different TP Size
#1470 opened Mar 12, 2025 by xiaojunjie Loading…
fix(moe): the missing argument 'router_dtype' of _DeepepManager.__init__
#1463 opened Mar 11, 2025 by AsakusaRinne Loading…
Draft: Youngeun/a2a hiding
#1460 opened Mar 10, 2025 by lhb8125 Loading…
[ENHANCEMENT] add z-loss (improved version)
#1442 opened Feb 28, 2025 by wdevazelhes Loading…
Replace deprecated numpy.product with numpy.prod to ensure compatibility with NumPy >=2.0
#1440 opened Feb 27, 2025 by mustious Loading…
1
fix seq_aux_loss for DeepSeek-V3
#1439 opened Feb 27, 2025 by yzlnew Loading…
1
fix a bug in load balancing loss aggregation when recompute is turned on
#1433 opened Feb 26, 2025 by lyuwen Loading…
a proof of concept for Distributed Muon
#1428 opened Feb 24, 2025 by toothacher17 Loading…
47
fix: return float instead of tensor from get_rotary_seq_len
#1419 opened Feb 20, 2025 by jasonchiu-codeium Loading…
Fix document regarding GQA (--group-query-attention) argument stale No activity in 60 days on issue or PR
#1401 opened Feb 12, 2025 by eagle705 Loading…
1
Fix issue in converting Mixtral 8x7B checkpoints from HF to MCore and update doc stale No activity in 60 days on issue or PR
#1397 opened Feb 11, 2025 by yeahdongcn Loading…
1
ProTip! Mix and match filters to narrow down what you’re looking for.