Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6511 commits
Select commit Hold shift + click to select a range
b5d50cb
Add simple GRPO functional test (#3323)
tdene Feb 11, 2026
1c245c7
ci: Concurrency control for merge-queue (#3353)
ko3n1g Feb 11, 2026
d9f075c
ci: Update golden value download script to work with Github (#3335)
chtruong814 Feb 11, 2026
d0b768f
Removing etc from main index page, shifted name of discussions (#3271)
megnvidia Feb 11, 2026
7d1acf6
fix: correct typos 'seperated' and 'recieved' (#3305)
thecaptain789 Feb 11, 2026
2807a4e
Improved PyTorch profiler and added PyTorch execution trace (#3273)
shengf-nv Feb 11, 2026
1959739
build: Bump TE on 2.12 (#3371)
ko3n1g Feb 11, 2026
f06e669
ci(hotfix): job conditions (#3376)
ko3n1g Feb 11, 2026
6467cbc
Update copy-pr-bot.yaml [skip ci]
github-actions[bot] Feb 12, 2026
faced51
Record moe routing decisions during inference. (#3034)
sidsingh-nvidia Feb 11, 2026
dedb6dd
[Main] Fix EP Overlap Bugs for Full-Iter CG (#3164)
Wohox Feb 12, 2026
11a4659
Avoid direct pickle import (#3375)
maanug-nv Feb 12, 2026
fe9279e
Delete old pretrain_* files (#3359)
Phlip79 Feb 12, 2026
7df15bd
Add Qwen3-VL support with Megatron-FSDP (#2841)
xuwchen Feb 12, 2026
c65fb25
Refactor Mamba chunked prefill (#3265)
santhnm2 Feb 12, 2026
47938af
Improved parallel logging of learning rate (#3319)
jstjohn Feb 12, 2026
a51c1c8
Add enhanced event tracking with TTFT measurement and compact seriali…
lmcafee-nvidia Feb 12, 2026
dbc444f
Add assertion that max_requests is divisible by tp_size (#3304)
santhnm2 Feb 12, 2026
1123fc0
Move to using the Inference OpenAI API server (#3107)
ArEsKay3 Feb 12, 2026
4184bfa
Revert "Move to using the Inference OpenAI API server (#3107)"
ko3n1g Feb 13, 2026
9119fae
Update moe github test cases. (#3077)
Victarry Feb 13, 2026
e0aa16b
Revert "Update moe github test cases. (#3077)"
ko3n1g Feb 13, 2026
28ccdaa
Split layer_specs to return Submodules instead of ModuleSpecs (#3255)
nschank Feb 13, 2026
76a9f47
ci: Remove gpu sanity check (#3420)
chtruong814 Feb 13, 2026
d10eb6f
[Critical-Bug] Fix Uneven PP for Mamba models (Nemotron3-nano) (#3399)
kevalmorabia97 Feb 13, 2026
2611830
Fix for rl (#3390)
shanmugamr1992 Feb 13, 2026
4578ed8
Add check for full_iteration scope before instantiating CudaGraphMana…
vasunvidia Feb 13, 2026
698feec
Fix broken links throughout (#3230)
megnvidia Feb 13, 2026
d401490
Extract intermediate embeddings of transformer block (#3060)
sajadn Feb 13, 2026
545bff9
Update copy-pr-bot.yaml [skip ci]
github-actions[bot] Feb 14, 2026
b890099
Decouple topk and loss from DSA Indexer (#3248)
kunlunl Feb 13, 2026
7a8f305
Move to using the Inference OpenAI API server (bis) (#3395)
tdene Feb 14, 2026
4beb8ca
Make Mamba inference state memory ratio configurable (#3322)
santhnm2 Feb 16, 2026
cbb47c8
Fix configs for RL model environments (#3441)
tdene Feb 16, 2026
8f1c2f8
Replace pickle with json in rl_utils (#3351)
tdene Feb 16, 2026
057c804
fix: correct typo in demo training example (#3428)
dndnda Feb 17, 2026
b218e64
Clean up logging inside inference flask server (#3437)
tdene Feb 17, 2026
3c69780
ci: Update release-docs workflow to use FW-CI-templates v0.72.0 (#3438)
chtruong814 Feb 17, 2026
74ef64e
Fix --tokenizer-hf-include-special-tokens (#3422)
jon-barker Feb 17, 2026
267cf1f
Update num_tokens_to_generate default for Gym (#3453)
tdene Feb 17, 2026
0627623
Fix slowdown in inference flask server (#3445)
tdene Feb 17, 2026
a22c40e
Add a normalized scale for MTP per token loss (#3159)
BestJuly Feb 17, 2026
d7500d4
[Bugfix] Fix nan loss caused by zero token in MTP (#3396)
BestJuly Feb 17, 2026
ad5a627
ci: Add testing branches
ko3n1g Feb 18, 2026
cd71d4c
chore: rotate oncall schedule
github-actions[bot] Feb 18, 2026
f1908bc
Log RL metrics per environment (#3446)
yobibyte Feb 18, 2026
1106df4
Move tensor offload/onload out of RL code (#3029)
tdene Feb 18, 2026
0672477
Add Engine event to the follow up requests after checkpointing (#3473)
ArEsKay3 Feb 18, 2026
7b016be
Fix another inference flask / Gym interaction (#3467)
tdene Feb 18, 2026
acb7273
adding in copyright blurb at the top of md file (#3394)
megnvidia Feb 18, 2026
fdde15e
[Megatron-FSDP] Add fsdp_all_gather_in_start_param_sync option in DDP…
shjwudp Feb 18, 2026
77f22f2
ci: Update release workflow to include changelog and publish docs (#3…
chtruong814 Feb 18, 2026
1666b45
ci(fix): Weekly GPT tests (#3443)
ko3n1g Feb 18, 2026
0d0943c
ci: Remove environments (#3462)
ko3n1g Feb 19, 2026
d07f16c
update HF tokenizer defaults (#3440)
dimapihtar Feb 19, 2026
1d694c2
PTQ changes for upcoming QAD (#3124)
AAnoosheh Feb 19, 2026
655cc8e
ci: Bump preflight to detect our svc (#3494)
ko3n1g Feb 19, 2026
7f35af4
build: Drop Python 3.10 support and pip install one-logger (#3485)
ko3n1g Feb 19, 2026
2d06cc9
ci: Bump pre-flight for Bot SSO (#3497)
ko3n1g Feb 19, 2026
a781f3c
Update copy-pr-bot.yaml [skip ci]
github-actions[bot] Feb 19, 2026
50ebe8e
Revert "build: Drop Python 3.10 support and pip install one-logger (#…
ko3n1g Feb 19, 2026
c191ae8
Fix chunked prefill edge cases (#3404)
santhnm2 Feb 19, 2026
9f611b7
ci: Enable MBridge downstream testing via PR (#3483)
ko3n1g Feb 19, 2026
7b6e226
ci: Remove gitlab docs build job and set LTS integration and function…
chtruong814 Feb 19, 2026
31bd4a3
[OMNIML-3232] ModelOpt: add full TE spec option and wire Mamba stack …
yueshen2016 Feb 20, 2026
9ba248e
Track off-policyness across RL steps (#3030)
tdene Feb 20, 2026
9d72f63
chore(beep boop 🤖): Bump (main) (2026-02-20)
github-actions[bot] Feb 20, 2026
b7aa6a0
ci: MBridge testing branch name during merge-queues (#3513)
ko3n1g Feb 20, 2026
7a36263
ci: Enable Dependabot Automerge (#3487)
ko3n1g Feb 20, 2026
e8fd432
ci: Also sync direct teams (#3484)
ko3n1g Feb 20, 2026
01b361c
Multimodal: fix argument checking (#3449)
faradawn Feb 20, 2026
773c113
Fix Megatron-FSDP optimizer state DCP checkpointing, and fix DTensor …
cspades Feb 20, 2026
b555baf
Update copy-pr-bot.yaml [skip ci]
github-actions[bot] Feb 21, 2026
32efeff
Renable full_iteration cuda graphs for inference. Add them for the ma…
sidsingh-nvidia Feb 20, 2026
a6d6dc6
do not add EoD (#3526)
arendu Feb 21, 2026
7124748
chore(beep boop 🤖): Bump (main) (2026-02-23)
github-actions[bot] Feb 23, 2026
3d1a4ba
Do not Slack notify for draft PRs (#3536)
Phlip79 Feb 23, 2026
9c6b69b
Update copy-pr-bot.yaml [skip ci]
github-actions[bot] Feb 23, 2026
6159399
remove deprecated SampleListWebdataset (#3407)
dimapihtar Feb 23, 2026
4aeeca8
remove deprecated get_te_version (#3413)
dimapihtar Feb 23, 2026
33df24c
remove deprecated async_grad_allreduce param (#3412)
dimapihtar Feb 23, 2026
eb0da52
remove deprecated mamba params (#3411)
dimapihtar Feb 23, 2026
82cbd82
remove deprecated params from model parallel config (#3408)
dimapihtar Feb 23, 2026
a0b9c16
Remove redundant CUDA calls in the LLaVA dataloader (#3476)
duncanriach Feb 23, 2026
fde3b90
Inference: Create finer grained cuda-graphs with better coverage of s…
sidsingh-nvidia Feb 23, 2026
23dd639
fix: skip non-tensor optimizer state entries in distrib_optimizer sav…
ahmadki Feb 23, 2026
2ec295e
remove is_unitialized & get_data_modulo_expert_parallel_group (#3414)
dimapihtar Feb 24, 2026
abf97eb
remove deprecated TE module (#3409)
dimapihtar Feb 24, 2026
7c7c9e1
remove encoder_and_decoder from enums (#3406)
dimapihtar Feb 24, 2026
dd39eb5
chore(beep boop 🤖): Bump (main) (2026-02-24)
github-actions[bot] Feb 24, 2026
cb24802
Add knobs to choose process groups for fully-parallel-save / load and…
sbak5 Feb 24, 2026
5f5f465
Fix off-by-2 error in RL sequence packing (#3551)
tdene Feb 24, 2026
a8efd34
Skip unnecessary flattening for Save / Load Planner (#3263)
sbak5 Feb 24, 2026
5dc98a6
Multimodal: fix model provider (#3508)
faradawn Feb 24, 2026
76b200c
docs: Enable nightly docs publish (#3546)
chtruong814 Feb 24, 2026
f721069
Ensure type-checker understands use of Submodules in unit tests (#3425)
nschank Feb 24, 2026
782e54b
Use copy_signature to preserve typing of pass-through methods (#3419)
nschank Feb 24, 2026
3597312
Ensure type-checker understands use of Submodules in MTP (#3308)
nschank Feb 24, 2026
44e27d0
Add mxfp8 quantization for inference linear layers (#3447)
santhnm2 Feb 24, 2026
2cac78b
Add single-process checkpoint save to avoid forked multiprocessing (#…
sbak5 Feb 24, 2026
08857d9
Fixed fp32 residuals (#3504)
mkhona-nvidia Feb 24, 2026
b17248a
Move config src files into a dedicated dir (#3570)
maanug-nv Feb 25, 2026
9bb2375
chore: rotate oncall schedule
github-actions[bot] Feb 25, 2026
0dba6a0
Revert "remove encoder_and_decoder from enums (#3406)" (#3579)
ko3n1g Feb 25, 2026
b359981
Fix default cuda graph persist arg. Add persist to rl common.sh. (#3584)
yobibyte Feb 25, 2026
60a25aa
Optimize away add request overheads in dummy ep cuda-graphed forward …
sidsingh-nvidia Feb 25, 2026
e7d80bd
ci: Test docs build (#3583)
ko3n1g Feb 25, 2026
80ef2ae
ci: Fix docs build for release (#3597)
ko3n1g Feb 26, 2026
2561a59
ci: Remove secrets (#3598)
ko3n1g Feb 26, 2026
19444b9
ci: Define secrets (#3599)
ko3n1g Feb 26, 2026
a08ad25
ci: gh-release-from-tag (#3600)
ko3n1g Feb 26, 2026
ac449c4
Ko3n1g/ci/remove twine username (#3601)
ko3n1g Feb 26, 2026
9088d4f
Add training code to MCore wheel (#3573)
maanug-nv Feb 26, 2026
dd4c0b7
FP8 attention knob for nvFP4 recipe (#3363)
vasunvidia Feb 26, 2026
c2782fd
Fix error with --load-main-params-from-ckpt (#3569)
guyueh1 Feb 26, 2026
c921891
ci: Create comment (#3610)
ko3n1g Feb 26, 2026
9100119
ci: Skip cleanup-taint-node jobs during deployments (#3612)
ko3n1g Feb 26, 2026
6161f7a
ci: No comment for release workflow (#3615)
ko3n1g Feb 26, 2026
36d8a9d
ci: Re-add release tag prefix (#3619)
ko3n1g Feb 26, 2026
1b1f5c4
docs: Fix version picker urls (#3621)
chtruong814 Feb 26, 2026
027e0f3
ci: Increase changelog generation max PRs fetched (#3620)
chtruong814 Feb 26, 2026
18c94d2
Add debug info to an assert. (#3588)
yobibyte Feb 26, 2026
e1a9ac9
fix: async_utils: explicit GC in persistent checkpoint worker loop (#…
sbak5 Feb 26, 2026
5f668c1
Fix: Perform sigmoid calculation in fp32 for aux loss stability (#2765)
CodersAcademy006 Feb 26, 2026
d3c10df
fix: forward use_te_activation_func flag in non-MoE GPT layer spec (#…
saakshigupta2002 Feb 26, 2026
7e3f670
Multimodal: Limit transformer version in Dockerfile (#3448)
faradawn Feb 26, 2026
6c0b9c6
Track and plot per-token off-policy in RL (#3515)
tdene Feb 26, 2026
a7c207f
Revert "Add single-process checkpoint save to avoid forked multiproce…
ko3n1g Feb 26, 2026
6503bf8
Multimodal: fix VQA dataset selection (#3464)
faradawn Feb 26, 2026
2f549e5
Multimodal: Fix multimodal training example - tokenizer, Triton Cache…
faradawn Feb 26, 2026
afbce84
Support TP > GQA for inference (#3627)
santhnm2 Feb 26, 2026
310082a
μP: Maximal Update Parameterization (#3058)
plugyawn Feb 26, 2026
7418b1b
Add flexible virtual pipeline parallel (fVPP) to hybrid model (#3377)
duncanriach Feb 26, 2026
53c5973
Explicitly close and join Pool in preprocess_data.py (#3592)
weijiac0619 Feb 27, 2026
36a95ae
remove indexer (#3416)
dimapihtar Feb 27, 2026
f0519b7
Multimodal: add load weights only (#3452)
faradawn Feb 27, 2026
61a293d
Add single-process checkpoint save to avoid forked multiprocessing (#…
sbak5 Feb 27, 2026
6287e7f
Update oncall schedule (#3632)
Phlip79 Feb 27, 2026
93d2739
Update copy-pr-bot.yaml [skip ci]
github-actions[bot] Feb 28, 2026
e8fe068
M-FSDP: Cancel erroneous grad accumulation check (#3629)
shjwudp Mar 2, 2026
107f6ae
chore(beep boop 🤖): Bump (main) (2026-03-02)
github-actions[bot] Mar 2, 2026
c7be214
Fix MoE aux loss tracker hang with MTP enabled (#3401)
Victarry Mar 2, 2026
044f1e3
Fix test data preparation (#3652)
janEbert Mar 2, 2026
2f8c9bc
Add GPTOSS Example with Megatron-LM + Megatron Bridge (#3018)
faradawn Mar 2, 2026
63cd60b
Add thd unit test main (#3617)
kunlunl Mar 2, 2026
c9312e6
Inference | KV prefix caching. (#3063)
lmcafee-nvidia Mar 2, 2026
b969f76
[Megatron-FSDP] Add dtype customization to Megatron-FSDP. (#3067)
cspades Mar 2, 2026
0810e63
CachedMetadataFileSystemReader: shared cache (#3326)
sbak5 Mar 2, 2026
7d1c016
Inference Optimized MoEs (#3496)
sidsingh-nvidia Mar 3, 2026
98495af
Log torch_memory_saver offload/onload (#3567)
tdene Mar 3, 2026
6fc7690
Prefix caching | Mamba memory only. (#3657)
lmcafee-nvidia Mar 3, 2026
9b18de4
Prefix caching | Coordinator scheduling. (#3665)
lmcafee-nvidia Mar 3, 2026
7dee32a
Adding manual Claude reviewer (#3679)
Phlip79 Mar 3, 2026
2570947
Nemo-RL Refit (#3520)
wdykas Mar 3, 2026
fa93d79
Add extra permissions and make other changes (#3683)
Phlip79 Mar 4, 2026
77a00ec
Claude should always comment something (#3685)
Phlip79 Mar 4, 2026
2caa681
[Cleanup] Remove the deprecated GroupedMLP (#3410)
dimapihtar Mar 4, 2026
470c6ea
chore: rotate oncall schedule
github-actions[bot] Mar 4, 2026
2a931a3
Fix illegal memory access with mamba inference (#3631)
tdene Mar 4, 2026
bb31e93
Fix illegal memory access with mamba inference (bis) (#3696)
tdene Mar 4, 2026
4fa9b5a
remove duplicate rerun_state_machine.set_mode(rerun_mode) (#3279)
YangWang92 Mar 4, 2026
d3a8584
Correct indexing when cp_comms_type is a list (#3389)
jeromeku Mar 4, 2026
bfd160b
Fix optional chat_completions returnables (#3519)
tdene Mar 4, 2026
f84e84e
ci: Claude code review (#3704)
ko3n1g Mar 4, 2026
bd1406f
ci: Fix event payload (#3705)
ko3n1g Mar 4, 2026
33476ff
ci: Use issue number (#3706)
ko3n1g Mar 4, 2026
fd21af4
ci: Finalize Claude review (#3707)
ko3n1g Mar 4, 2026
b1b4df6
ci: Add codecov yml (#3455)
thomasdhc Mar 4, 2026
7ea354b
Robust signaling for coordinator inference (#3563)
tdene Mar 4, 2026
f91c4bb
Fix memory issue in mxfp8 model init (#3461)
WanZzzzzz Mar 4, 2026
c5d8f6b
Update copy-pr-bot.yaml [skip ci]
github-actions[bot] Mar 5, 2026
fa6063d
adding public_docs_features: True to get proper legal footer… (#3681)
megnvidia Mar 4, 2026
a2381d8
add --overlap-param-gather support for layer-wise optimizer. lots of …
mchrzanowski Mar 5, 2026
657d33b
ci: Mount and enforce HF_HOME (#3700)
ko3n1g Mar 5, 2026
da47e64
Add flags for changing Mamba inference state tensor dtype (#3660)
santhnm2 Mar 5, 2026
94a903b
chore: CLI launch internal CI (#3695)
ko3n1g Mar 5, 2026
485428f
Change Review Process (#3659)
Phlip79 Mar 5, 2026
d1cce0c
ci: Separate queues for internal/external contributors (#3718)
ko3n1g Mar 5, 2026
6bd5c12
Update to correct token (#3724)
Phlip79 Mar 5, 2026
f5b2ec0
build: Bump to NGC PyTorch 26.02 (#3474)
ko3n1g Mar 5, 2026
bb84493
Claude: use Opus 4.6 and auto-review on ready (#3727)
Phlip79 Mar 6, 2026
41daf81
Claude to add complexity label (#3709)
Phlip79 Mar 6, 2026
0d42bc6
Offload Flask frontend to separate process (#3648)
santhnm2 Mar 6, 2026
d3528a2
fix(moe): fix TE general_gemm API change (#3582)
hxbai Mar 6, 2026
43df309
Review process fixes (#3728)
Phlip79 Mar 6, 2026
bde8264
ci: Update golden values after PyT bump (#3733)
ko3n1g Mar 6, 2026
a979332
chore: Use PAT for CLI Launcher (#3734)
ko3n1g Mar 6, 2026
bb451db
Print more verbose error message about incorrect `model_parallel_size…
rj42 Mar 6, 2026
de63aa8
ci: Add missing gitlab rule (#3735)
ko3n1g Mar 6, 2026
37ca715
[main] Add TE CUDA Graph Support for Vision Encoder (#3293)
tomlifu Mar 6, 2026
17de0db
Optimize process management and delete operations for async save (#3262)
sbak5 Mar 6, 2026
6ec369d
Align gpt-oss window-size with 128-token sliding window (#2771)
returnL Mar 6, 2026
26f9444
fix: temperature validation error message 1000.0 -> 100.0 (#2688)
CreeperLKF Mar 6, 2026
e19fbe2
RL: Hybrid MoE training cudagraphs and fix training <-> inference tra…
mathemakitten Mar 6, 2026
c1e675f
Fix dynamic inference and GRPO functional tests (#3740)
santhnm2 Mar 6, 2026
0cfa420
Swap oncall (#3585)
janEbert Mar 6, 2026
932b767
Update copy-pr-bot.yaml [skip ci]
github-actions[bot] Mar 7, 2026
b09ee64
[bugfix] fix the bug that loss: 0 will not be printed (#1555)
leisuzz Mar 9, 2026
8318b80
Fused dLN + add in backwards pass (#3384)
CarlosGomes98 Mar 9, 2026
597721a
chore(beep boop 🤖): Bump (main) (2026-03-09)
github-actions[bot] Mar 9, 2026
116a7fa
Claude: run actions on target branch (#3745)
Phlip79 Mar 9, 2026
56158bb
revert of #2658 (#3736)
dimapihtar Mar 9, 2026
452fc11
Update README Quick Start (#3596)
ilml Mar 9, 2026
d904a68
Re-enable tests which were failing on #3373 (#3757)
mathemakitten Mar 9, 2026
94ff0dc
Check reviews properly (#3756)
Phlip79 Mar 9, 2026
ce66b22
Update copy-pr-bot.yaml [skip ci]
github-actions[bot] Mar 10, 2026
0e19bf1
Add CP + Sequence Packing support for Mimo (#2135)
mehraakash Mar 10, 2026
fca1679
MXFP8 refit (#3742)
wdykas Mar 10, 2026
397772e
Claude: update token usage (#3760)
Phlip79 Mar 10, 2026
3eea580
Handle Tool Call Argument Parsing (#3662)
sancha Mar 10, 2026
e970199
RL support for nanov3 sft checkpoint (#3741)
jon-barker Mar 10, 2026
ba497c9
add mix_hidden_states option in conversion (#3655)
yeyu-nvidia Mar 10, 2026
22c69fa
ci: Optimize release-configs for GB200 (#3541)
ko3n1g Mar 10, 2026
0f47a1a
Update copy-pr-bot.yaml [skip ci]
github-actions[bot] Mar 11, 2026
204e7d5
Add absorbed-mla (#3198)
kunlunl Mar 10, 2026
f544034
feat(checkpoint): zero-copy storage sharing in CheckpointWithoutOutpu…
Victarry Mar 11, 2026
8fd390d
Fuse MLA DOWN projection GEMMs (#3039)
cjld Mar 11, 2026
7d52694
fix: skip FSDP DTensor boundary validation under fake process group (…
Victarry Mar 11, 2026
16a8cdb
[main] fix(moe): fix the bug where gate was not sliced when kv_head <…
yuzhongw-nvidia Mar 11, 2026
b8e23d5
fix(offload): reset activation offload manager after eval as well as …
rapatel Mar 11, 2026
07e512a
chore: rotate oncall schedule
github-actions[bot] Mar 11, 2026
e20b89a
Improve error logging when invalid number of tokens is requested. (#3…
yobibyte Mar 11, 2026
5bc89f3
Add NVIDIA-Nemotron-3-Super-120B-A12B-BF16 to ModelOpt examples (#3805)
jenchen13 Mar 11, 2026
da46946
build: Bump TE2.13 (#3800)
ko3n1g Mar 11, 2026
8e64e69
Ensure dummy_forward does not attempt to run cudagraphs (#3789)
jalbericiola Mar 11, 2026
8f539df
Add speculative decoding support with MTP layers (#3594)
santhnm2 Mar 11, 2026
39472d8
Shanmugamr1992/megatron inference ultra (#3784)
shanmugamr1992 Mar 11, 2026
d997820
Fix backward compatibility issue with MFSDP `--grad-reduce-in-bf16` (…
shjwudp Mar 12, 2026
251a754
feat: add NCCL flight recorder configuration support (#3806)
sbak5 Mar 12, 2026
5a3aa17
Revert "Ensure dummy_forward does not attempt to run cudagraphs (#378…
ko3n1g Mar 12, 2026
5b25326
Fix if statement in main (#3833)
tdene Mar 12, 2026
6657173
Update golden values of weekly tests (#3829)
ko3n1g Mar 12, 2026
4736aed
build: Loosen TE restriction (#3827)
ko3n1g Mar 12, 2026
1d5e68b
Upgrade GitHub Actions for Node 24 compatibility (#3830)
ko3n1g Mar 12, 2026
46227e0
Do not let chunked prefill generate decode logprobs (#3777)
tdene Mar 12, 2026
e08dc9d
Prevent double serialization inside Flask server (#3653)
tdene Mar 12, 2026
29e798a
Allow RL to run inference-only via skip-train (#3744)
tdene Mar 12, 2026
4c1d0e4
Announce Python 3.12 migration (#3825)
ko3n1g Mar 12, 2026
fabbcdf
Update copy-pr-bot.yaml [skip ci]
github-actions[bot] Mar 13, 2026
b250472
ci: Skip test_wrong_cuda_graph_impl_returns_false in LTS (#3847)
chtruong814 Mar 13, 2026
7ca9dc5
ci: Mark TestCoordinator.test_throughput as flaky (#3849)
chtruong814 Mar 13, 2026
f261906
find optimal number of workers (#3699)
dimapihtar Mar 13, 2026
b7437fe
remove encoder_and_decoder (#3836)
dimapihtar Mar 13, 2026
8a806e5
ci: Skip more tests in test_vision_cuda_graphs for LTS (#3860)
chtruong814 Mar 13, 2026
90fee1f
Ensure that inference dummy_forward does not try to match on a cudagr…
mathemakitten Mar 13, 2026
d1b8e27
Add unit tests for speculative decoding (#3817)
santhnm2 Mar 13, 2026
d4ac04f
Fix flakiness due to timing between shutdowns (#3857)
tdene Mar 13, 2026
87eb3c2
Exposing interleave argument for fused_apply_rotary_pos_emb_thd (#3794)
huvunvidia Mar 13, 2026
9a19203
ci: install nvidia-resiliency-ext from source (#3861)
ko3n1g Mar 13, 2026
f8becec
Miscellaneous inference bug fixes (#3840)
santhnm2 Mar 14, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
39 changes: 39 additions & 0 deletions .coderabbit.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,39 @@
# yaml-language-server: $schema=https://coderabbit.ai/integrations/schema.v2.json
language: "en-US"

# Only comment on Critical/Major bugs. No Minor, Trivial, or style comments.
tone_instructions: "Only comment on Critical or Major bugs. Never comment on Minor issues, style, refactoring, or suggestions. When in doubt, stay silent."

reviews:
# Use chill profile - filters out nitpicks automatically
profile: "chill"

# Disable all summary features
high_level_summary: false
high_level_summary_in_walkthrough: false

# Disable walkthrough comment entirely
collapse_walkthrough: true
changed_files_summary: false
sequence_diagrams: false

# Disable status/effort estimates
review_status: false
commit_status: false
estimate_code_review_effort: false

# Disable auto-suggestions for labels/reviewers
suggested_labels: false
suggested_reviewers: false

# Disable related issues/PRs lookup
assess_linked_issues: false
related_issues: false
related_prs: false

# Auto-review disabled - only review when explicitly requested via @coderabbitai review
auto_review:
enabled: false

chat:
auto_reply: true
5 changes: 0 additions & 5 deletions .coveragerc

This file was deleted.

4 changes: 4 additions & 0 deletions .flake8
Original file line number Diff line number Diff line change
@@ -0,0 +1,4 @@
[flake8]
max-line-length = 100
extend-ignore = E203,E501,F401,E402,E714
per-file-ignores = __init__.py:F401
61 changes: 61 additions & 0 deletions .github/CODEOWNERS
Original file line number Diff line number Diff line change
@@ -0,0 +1,61 @@
megatron/core/ @NVIDIA/core-adlr @NVIDIA/core-nemo

megatron/core/models/gpt/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/gpt

megatron/core/models/multimodal/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/multi-modal

megatron/core/models/mamba/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/hybrid-mamba
megatron/core/ssm/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/hybrid-mamba

megatron/core/datasets/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/datasets

megatron/core/tokenizers/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/tokenizers

megatron/core/distributed/fsdp/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/megatron-fsdp

megatron/core/transformer/fsdp_dtensor_checkpoint.py @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/megatron-fsdp

megatron/core/dist_checkpointing/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/dist-checkpointing

megatron/core/optimizer/distrib_optimizer/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/dist-optimizer

megatron/core/inference/modelopt_support @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/quantization-and-inference

megatron/core/datasets/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/datasets

megatron/core/pipeline_parallel/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/pipeline-parallelism

megatron/core/transformer/ @NVIDIA/core-adlr @NVIDIA/core-nemo

megatron/core/transformer/moe/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/mixture-of-experts-adlr @NVIDIA/mixture-of-experts-devtech

megatron/core/inference/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/inference

megatron/core/parallel_state.py @NVIDIA/core-adlr @NVIDIA/core-nemo

megatron/core/post_training/ @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/post-training

megatron/post_training/ @NVIDIA/post-training

megatron/core/transformer/cuda_graphs.py @NVIDIA/core-adlr @NVIDIA/core-nemo @NVIDIA/cuda-graphs

.gitlab/ @NVIDIA/ci
.github/ @NVIDIA/ci
.gitlab-ci.yml @NVIDIA/ci
docker/ @NVIDIA/ci
tests/functional_tests/python_test_utils/ @NVIDIA/ci
tests/functional_tests/shell_test_utils/ @NVIDIA/ci
tests/test_utils/recipes/ @NVIDIA/ci
tests/unit_tests/run_ci_test.sh @NVIDIA/ci

# API Backwards Compatibility Check
scripts/check_api_backwards_compatibility.py @NVIDIA/ci
scripts/README_API_COMPAT.md @NVIDIA/ci
.github/workflows/check_api_backwards_compatibility_workflow.yml @NVIDIA/ci
docs/api-backwards-compatibility-check.md @NVIDIA/ci
tests/unit_tests/test_api_backwards_compat_setup.py @NVIDIA/ci

megatron/rl/ @NVIDIA/reinforcement-learning
examples/rl/ @NVIDIA/reinforcement-learning
test/unit_tests/test_rl_utils.py @NVIDIA/reinforcement-learning
train_rl.py @NVIDIA/reinforcement-learning
29 changes: 29 additions & 0 deletions .github/ISSUE_TEMPLATE/bug_report.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,29 @@
---
name: Bug report
about: Create a report to help us improve the repository or project
title: ""
labels: bug
assignees: ''

---

**Describe the bug**

A clear and concise description of what the bug is. Tag the [@mcore-oncall](https://github.com/orgs/NVIDIA/teams/mcore-oncall)
to get oncall's attention to this issue.

**Steps/Code to reproduce bug**

Please list *minimal* steps or code snippet for us to be able to reproduce the bug.

A helpful guide on on how to craft a minimal bug report http://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports.


**Expected behavior**

A clear and concise description of what you expected to happen.


**Additional context**

Add any other context about the problem here.
2 changes: 2 additions & 0 deletions .github/ISSUE_TEMPLATE/config.yml
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
blank_issues_enabled: false

23 changes: 23 additions & 0 deletions .github/ISSUE_TEMPLATE/feature_request.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,23 @@
---
name: Feature request
about: Suggest an idea for this project
title: ""
labels: enhancement
assignees: ''

---

**Is your feature request related to a problem? Please describe.**
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

Tag the [@mcore-oncall](https://github.com/orgs/NVIDIA/teams/mcore-oncall)
to get oncall's attention to this issue.

**Describe the solution you'd like**
A clear and concise description of what you want to happen.

**Describe alternatives you've considered**
A clear and concise description of any alternative solutions or features you've considered.

**Additional context**
Add any other context or screenshots about the feature request here.
13 changes: 13 additions & 0 deletions .github/ISSUE_TEMPLATE/question.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,13 @@
---
name: QUESTION
about: Ask a question about Megatron-LM that is not a bug, regression or enhancement
request
title: "[QUESTION]"
labels: ''
assignees: ''

---

**Your question**
Ask a clear and concise question about Megatron-LM. Tag the [@mcore-oncall](https://github.com/orgs/NVIDIA/teams/mcore-oncall)
to get oncall's attention to this issue.
40 changes: 40 additions & 0 deletions .github/ISSUE_TEMPLATE/regression.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,40 @@
---
name: REGRESSION
about: Report a regression in speed or accuracy due to a Megatron-LM update
title: "[REGRESSION]"
labels: ''
assignees: ''

---

**Describe the regression**
A clear and concise description of what the regression is. Tag the [@mcore-oncall](https://github.com/orgs/NVIDIA/teams/mcore-oncall)
to get oncall's attention to this issue.

**To Reproduce**
Steps to reproduce the behavior. The easier it is to reproduce the faster it will get maintainer attention.

**Previous performance**
What speed or accuracy did you previously see.

**New performance**
What speed or accuracy do you see after the update.

**Stack trace/logs**
If applicable, add the stack trace or logs related to the regression.

**Environment (please complete the following information):**
- Previous Megatron-LM commit ID
- New Megatron-LM commit ID
- Previous PyTorch version
- New PyTorch version
- Previous CUDA version
- New CUDA version
- Previous NCCL version
- New NCCL version

**Proposed fix**
If you have a proposal for how to fix the issue state it here or link to a PR.

**Additional context**
Add any other context about the problem here.
Loading