Skip to content
Navigation Menu
Sign in
Appearance settings
Platform
AI CODE CREATION
GitHub Copilot
Write better code with AI
GitHub Copilot app
Direct agents from issue to merge
MCP Registry
Integrate external tools
DEVELOPER WORKFLOWS
Actions
Automate any workflow
Codespaces
Instant dev environments
Issues
Plan and track work
Code Review
Manage code changes
Code Quality
Enforce quality at merge
APPLICATION SECURITY
GitHub Advanced Security
Find and fix vulnerabilities
Code security
Secure your code as you build
Secret protection
Stop leaks before they start
EXPLORE
Why GitHub
Documentation
Blog
Changelog
Marketplace
View all features
Solutions
BY COMPANY SIZE
Enterprises
Small and medium teams
Startups
Nonprofits
BY USE CASE
App Modernization
DevSecOps
DevOps
CI/CD
View all use cases
BY INDUSTRY
Healthcare
Financial services
Manufacturing
Government
View all industries
View all solutions
Resources
EXPLORE BY TOPIC
AI
Software Development
DevOps
Security
View all topics
EXPLORE BY TYPE
Customer stories
Events & webinars
Ebooks & reports
Business insights
GitHub Skills
SUPPORT & SERVICES
Documentation
Customer support
Community forum
Trust center
Partners
View all resources
Open Source
COMMUNITY
GitHub Sponsors
Fund open source developers
PROGRAMS
Security Lab
Maintainer Community
Accelerator
GitHub Stars
Archive Program
REPOSITORIES
Topics
Trending
Collections
Enterprise
ENTERPRISE SOLUTIONS
Enterprise platform
AI-powered developer platform
AVAILABLE ADD-ONS
GitHub Advanced Security
Enterprise-grade security features
Copilot for Business
Enterprise-grade AI features
Premium Support
Enterprise-grade 24/7 support
Pricing
Search
/
Sign in
Sign up
Appearance settings
You signed in with another tab or window.
Reload
to refresh your session.
You signed out in another tab or window.
Reload
to refresh your session.
You switched accounts on another tab or window.
Reload
to refresh your session.
Dismiss alert
{{ message }}
Uh oh!
There was an error while loading.
Please reload this page
.
NVIDIA
/
Megatron-LM
Public
Notifications
You must be signed in to change notification settings
Fork
4.4k
Star
17.5k
Code
Issues
393
Pull requests
804
Discussions
Actions
Projects
Security and quality
0
Insights
Additional navigation options
Code
Issues
Pull requests
Discussions
Actions
Projects
Security and quality
Insights
[dev]: Fix mHC boundaries in EP overlap schedule
- #5471
#5471
Merged
jingqiny-99
merged 24 commits into
NVIDIA:dev
NVIDIA/Megatron-LM:dev
from
jingqiny-99:jingqiny/fix-mhc-ep-overlap-dev
jingqiny-99/Megatron-LM:jingqiny/fix-mhc-ep-overlap-dev
Copy head branch name to clipboard
Jul 22, 2026
Conversation
Commits
24
(24)
Checks
Files changed
Merged
[dev]: Fix mHC boundaries in EP overlap schedule
#5471
jingqiny-99
merged 24 commits into
NVIDIA:dev
NVIDIA/Megatron-LM:dev
from
jingqiny-99:jingqiny/fix-mhc-ep-overlap-dev
jingqiny-99/Megatron-LM:jingqiny/fix-mhc-ep-overlap-dev
Copy head branch name to clipboard
Commits
Commits on Jul 19, 2026
Fix mHC boundaries in EP overlap schedule
Show description for 228c35f
jingqiny-99
committed
228c35f
View commit details
Copy full SHA for 228c35f
Browse repository at this point
Address mHC EP overlap review comments
Show description for 47f91ef
jingqiny-99
committed
47f91ef
View commit details
Copy full SHA for 47f91ef
Browse repository at this point
Address mHC EP overlap follow-up review
Show description for 6c2f9e1
jingqiny-99
committed
6c2f9e1
View commit details
Copy full SHA for 6c2f9e1
Browse repository at this point
Add configurable mHC overlap scheduling
Show description for 53d9070
jingqiny-99
committed
53d9070
View commit details
Copy full SHA for 53d9070
Browse repository at this point
Replace mHC profiling artifacts with a schedule timeline
Show description for 3835ced
jingqiny-99
committed
3835ced
View commit details
Copy full SHA for 3835ced
Browse repository at this point
Remove mHC high-priority stream scheduling
Show description for bd2d3e3
jingqiny-99
committed
bd2d3e3
View commit details
Copy full SHA for bd2d3e3
Browse repository at this point
Add compute-stream scheduling for mHC post
Show description for 2238278
jingqiny-99
committed
2238278
View commit details
Copy full SHA for 2238278
Browse repository at this point
Remove unused mHC recompute group bounds
Show description for 8dcd316
jingqiny-99
committed
8dcd316
View commit details
Copy full SHA for 8dcd316
Browse repository at this point
Fix mHC boundary and HybridMoE config tests
Show description for b40acc0
jingqiny-99
committed
b40acc0
View commit details
Copy full SHA for b40acc0
Browse repository at this point
Fix: docstring
Show description for 245cdea
jingqiny-99
committed
245cdea
View commit details
Copy full SHA for 245cdea
Browse repository at this point
Delete docs/images/mhc_overlap/mhc_ep_overlap_partial_cuda_graph_timeline.svg
Show description for b285936
jingqiny-99
committed
b285936
View commit details
Copy full SHA for b285936
Browse repository at this point
Finalize mHC decoder boundary at the terminal schedule node
Show description for dc54cc2
jingqiny-99
and
claude
committed
dc54cc2
View commit details
Copy full SHA for dc54cc2
Browse repository at this point
Restore activation dtype for dense mHC MLP; relax interleaved MTP grad tolerance
Show description for 8dcc183
jingqiny-99
and
claude
committed
8dcc183
View commit details
Copy full SHA for 8dcc183
Browse repository at this point
Fix dense mHC MLP dtype cast to target params_dtype
Show description for 93d03c1
jingqiny-99
and
claude
committed
93d03c1
View commit details
Copy full SHA for 93d03c1
Browse repository at this point
Remove mhc_post_on_compute_stream flag and the separate mHC-post node path
Show description for d2b9c9f
jingqiny-99
and
claude
committed
d2b9c9f
View commit details
Copy full SHA for d2b9c9f
Browse repository at this point
upd: fix lint
Show description for a9c6898
jingqiny-99
committed
a9c6898
View commit details
Copy full SHA for a9c6898
Browse repository at this point
Fix MTP delayed-wgrad callable selection under hyper-connections
Show description for f5cc9c9
jingqiny-99
and
claude
committed
f5cc9c9
View commit details
Copy full SHA for f5cc9c9
Browse repository at this point
Extend mHC EP-overlap tests: delayed wgrad, MTP recompute, full-iteration CUDA graph
Show description for 054ec0f
jingqiny-99
and
claude
committed
054ec0f
View commit details
Copy full SHA for 054ec0f
Browse repository at this point
Set params_dtype in PP mHC forward test config
Show description for b0c506f
jingqiny-99
and
claude
committed
b0c506f
View commit details
Copy full SHA for b0c506f
Browse repository at this point
Flush CSA and DSA deferred weight gradients through core_attention
Show description for 35206e8
jingqiny-99
and
claude
committed
35206e8
View commit details
Copy full SHA for 35206e8
Browse repository at this point
Add delayed-wgrad flush tests for CSA and DSA attention variants
Show description for 317727d
jingqiny-99
and
claude
committed
317727d
View commit details
Copy full SHA for 317727d
Browse repository at this point
Disambiguate mHC recompute NVTX labels and trim a stale comment
Show description for 2a8f4a7
jingqiny-99
and
claude
committed
2a8f4a7
View commit details
Copy full SHA for 2a8f4a7
Browse repository at this point
Commits on Jul 21, 2026
Merge remote-tracking branch 'origin/dev' into jingqiny/fix-mhc-ep-overlap-dev
Show description for dd14dc7
jingqiny-99
and
claude
committed
dd14dc7
View commit details
Copy full SHA for dd14dc7
Browse repository at this point
Commits on Jul 22, 2026
Merge branch 'dev' into jingqiny/fix-mhc-ep-overlap-dev
Show description for db98830
jingqiny-99
authored
db98830
View commit details
Copy full SHA for db98830
Browse repository at this point
You can’t perform that action at this time.
