Skip to content

From NVIDIA Megatron-LM for visibility#18

Open
RaymondLi0 wants to merge 7284 commits into
bigcode-project:multi-query-attentionfrom
NVIDIA:main
Open

From NVIDIA Megatron-LM for visibility#18
RaymondLi0 wants to merge 7284 commits into
bigcode-project:multi-query-attentionfrom
NVIDIA:main

Conversation

@RaymondLi0

@RaymondLi0 RaymondLi0 commented Jan 24, 2023

Copy link
Copy Markdown
Collaborator

No description provided.

@RaymondLi0 RaymondLi0 changed the base branch from multi-query-attention to before-merge June 20, 2023 20:12
@RaymondLi0 RaymondLi0 changed the base branch from before-merge to multi-query-attention June 20, 2023 20:12
Wohox and others added 27 commits May 27, 2026 18:10
@Wohox
[fix] Release MTP assertion when EP overlap with PP=1 (#4796)
286445c
@ko3n1g
fix(test): widen iter-time steady-state window for short tests (#5023)
0cb4034 
Signed-off-by: oliver könig <okoenig@nvidia.com>
@shanmugamr1992 @claude
Perf fix (#4996)
805e24e 
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@Phlip79
Add dev-feature preservation gate and change schedule (#4773)
457e3f7
@ko3n1g
chore(test): remove orphan nemotron3_super_release_g200 dir (#5024)
146d171 
Signed-off-by: oliver könig <okoenig@nvidia.com>
@Phlip79
Ignore Claude worktree directory (#5020)
7be1748
@github-actions
Update copy-pr-bot.yaml [skip ci]
9d23a73
@balasaajay
ci: update CI workflow conditions for integration tests (#4658)
3e00820
@Phlip79
Add NVSkills CI request workflow (#5033)
0010683
@maanug-nv
DDP wrap pg size fixes (#5006)
2a4f820 
Signed-off-by: Maanu Grover <maanug@nvidia.com>
@Wohox
fix(layer_wise): tag MTP-stage word_embeddings as is_embedding_or_out…
e0afe89 
…put_parameter (#5034)
@balasaajay
Move LTS dependencies from pyproject.toml to Dockerfile.ci.lts (#4877)
4f2e7ce
@kevalmorabia97 @claude
Use shared ModelOpt calibration loop on 0.45+ with 0.44 fallback fix (#…
88efdf7 
…4881)

Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
@ko3n1g
test(release): skip golden comparison on intermediate resume windows (#…
a5c2d1b 
…5040)

Signed-off-by: oliver könig <okoenig@nvidia.com>
@liding-nv
[mimo] Thread position_ids through MimoModel for multimodal RoPE (#4938)
2ebca1e 
Signed-off-by: Li Ding <liding@nvidia.com>
@ko3n1g
build: Switch DSv3 on H100 to HybridEP (#5039)
e6ff4c3 
Signed-off-by: oliver könig <okoenig@nvidia.com>
@kevalmorabia97 @claude
Fix: Import unwrap_model from megatron.core.utils in modelopt examples (
ba9e0ea 
#5045)

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@YangFei1990 @claude
Simple and stable Inference APIs (#4697)
3c39d98 
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@balasaajay
ci: Add notification step for MBridge downstream test results (#5028)
f63a46e
@Phlip79
Delete output tensor early (#4742)
f1b5516
@sraman-rgb @sraman @chtruong814
Support ScaledSReLU in TE grouped MLP fuser (#4859)
f8e3885 
Co-authored-by: sraman <sraman@users.noreply.github.com>
Co-authored-by: Siddhartha Raman S <sraman@login-lyris01.lyris.clusters.nvidia.com>
Co-authored-by: a <a>
Co-authored-by: Charlie Truong <chtruong@nvidia.com>
Co-authored-by: sraman-rgb <270218152+sraman-rgb@users.noreply.github.com>
@yfw @gshennvm @claude
Skip gradient updates when grad norm exceeds threshold (#3460)
1801316 
Signed-off-by: Yi-Fu Wu <yifu.wu@gmail.com>
Co-authored-by: Gerald Shen <geshen@nvidia.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@Phlip79 @svc-nvskills-signing
Add 9 user skills (#5066)
6e0d14a 
Signed-off-by: nvskills-svc-account <svc-nvskills-signing@nvidia.com>
Co-authored-by: nvskills-svc-account <svc-nvskills-signing@nvidia.com>
@ko3n1g
test(nemotron): align nemotron3 super GB200 goldens with exit-interva…
54dc530 
…l 4768 (#5069)

Signed-off-by: oliver könig <okoenig@nvidia.com>
@balasaajay
chore: Update transformer-engine dependency to version 2.16.0 (#4992)
6e091e1
@maanug-nv
Update energon version requirement (#4572)
52d1d68 
Signed-off-by: Maanu Grover <maanug@nvidia.com>
@YangFei1990 @claude
Fix test failures for new inference APIs (#5068)
791a45f 
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
github-actions Bot and others added 30 commits June 29, 2026 09:14
@github-actions
Update copy-pr-bot.yaml [skip ci]
25f6a09
@asolergi-nv
[CI] Fix gpt_dynamic_inference_tp2_pp2_ep2_gptoss_20b_swa tests (#5527
522a9dd 
)

Signed-off-by: Antoni-Joan Solergibert <asolergibert@nvidia.com>
@lmcafee-nvidia
Implement async scheduling for dynamic inference (#5453)
3c327f3 
Signed-off-by: Lawrence McAfee <lmcafee@nvidia.com>
@ko3n1g
build: bump transformer-engine to release_v2.16.post (#5517)
b9f7fb6 
Signed-off-by: oliver könig <okoenig@nvidia.com>
@janEbert
Fix isort target Python version (#5567)
819912c 
Signed-off-by: janEbert <janpabloe@nvidia.com>
@janEbert
Fix PR template typo (#5566)
7b6eb02 
Signed-off-by: janEbert <janpabloe@nvidia.com>
@anlthms
Deduplicate tensor-splitting utility (#5545)
872442a 
Signed-off-by: Anil Thomas <anlthms@users.noreply.github.com>
@Phlip79 @ko3n1g
Add CI duties to oncall (#5510)
99b56a7 
Signed-off-by: Philip Petrakian <ppetrakian@nvidia.com>
Co-authored-by: oliver könig <okoenig@nvidia.com>
@yashaswikarnati
Thread dp_cp/expt_dp process groups through checkpoint load path (#5579)
817c1d5 
Signed-off-by: ykarnati <ykarnati@nvidia.com>
@rapatel
Fix TEGroupedMLP pre-backward unshard in fine-grained FSDP hooks for … (
f285ea5 
#4990)

Signed-off-by: Ritesh Patel <ripatel@nvidia.com>
@maanug-nv
[training migration] Finish ModelBuilder integration (#5516)
223e244 
Signed-off-by: Maanu Grover <maanug@nvidia.com>
@github-actions
Update copy-pr-bot.yaml [skip ci]
4849143
@Phlip79
Use NVIDIA inference credentials for Claude actions (#5589)
a3d761b 
Signed-off-by: Philip Petrakian <ppetrakian@nvidia.com>
@Phlip79
Update PR instructions (#5592)
36d11ce 
Signed-off-by: Philip Petrakian <ppetrakian@nvidia.com>
@Phlip79
Update mcore skill owners (#5586)
f972da7 
Signed-off-by: Philip Petrakian <ppetrakian@nvidia.com>
@github-actions
chore: rotate oncall schedule
8b3d8a5
@Phlip79
Add /claude fix workflow for on-demand PR fixes (#4862)
3c08255 
Signed-off-by: Philip Petrakian <ppetrakian@nvidia.com>
@github-actions
Update copy-pr-bot.yaml [skip ci]
4c4a8ee
@cspades
[Megatron-FSDP] MaxPoolAllocator for double-buffering hybrid architec…
adfe9e1 
…tures. (#5462)

Signed-off-by: Cory Ye <cye@nvidia.com>
@Pearblossom-M
fix(tensor_parallel): _reduce returns unreduced tensor for non-contig… (
60f338a 
#5338)

Signed-off-by: ma-jh <3364870135@qq.com>
@jenchen13 @realAsma @claude
Add Auto Quantize in ModelOpt quantize example (#4821)
2b551c6 
Signed-off-by: Jennifer Chen <jennifchen@nvidia.com>
Signed-off-by: realAsma <akuriparambi@nvidia.com>
Co-authored-by: realAsma <akuriparambi@nvidia.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@lhb8125 @yaox12
[Main][feat] Support CUDA Graph capture offloading modules (#3697)
69c4868 
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>
Signed-off-by: hongbinl <hongbinl@nvidia.com>
Signed-off-by: Hongbin Liu <lhb8125@users.noreply.github.com>
Co-authored-by: Xin Yao <xiny@nvidia.com>
@yashaswikarnati @claude
E2E heterogenous non colocated MiMo training (#5602)
0522099 
Signed-off-by: ykarnati <ykarnati@nvidia.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@jiemingz
Optimize memory usage of partial CUDA graphs (#5451)
5e4fe9b 
Signed-off-by: Jieming Zhang <jiemingz@nvidia.com>
Signed-off-by: root <jiemingz@nvidia.com>
@wujingyue
Document stacked dependent PR handling in split PR skill (#5496)
ee623a9 
Signed-off-by: Jingyue Wu <jingyuew@nvidia.com>
@Phlip79
Fix Claude reaction permissions (#5613)
060371c 
Signed-off-by: Philip Petrakian <ppetrakian@nvidia.com>
@balasaajay
Fix smoke BERT/T5 test failures (#5629)
6608a05 
Signed-off-by: Ajay Balasa <abalasa@nvidia.com>
@wujingyue @claude
Add NCCL symmetric-memory staging to experimental FSDP (#5440)
25f6117 
Signed-off-by: Jingyue Wu <jingyuew@nvidia.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@balasaajay
Add smoke test notification functionality and update notify script (#…
d89aae5 
…5631)

Signed-off-by: Ajay Balasa <abalasa@nvidia.com>
@balasaajay
Update golden value files for GPT-3 weekly (#5459)
4828d65 
Signed-off-by: Ajay Balasa <abalasa@nvidia.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.