Skip to content
Navigation Menu
Sign in
Products
Platform
GitHub
Repository hosting, code review, and collaboration
Copilot
AI coding agent for tasks, projects, and automated work
Build & Ship
Actions
CI/CD and automation
Code Quality
Quality analysis, fixes, and enforcement
Codespaces
Cloud development environments
Secure
Code Security
Vulnerability detection and remediation
Secret Protection
Secret scanning and leak prevention
All Features
Developers
Learn
Documentation
Training
Certifications
Build
Build with Copilot
Customize Copilot
Integrations
Explore
Trending repos
GitHub Sponsors
Developer articles
Changelog
Changelog
Your source for the latest features, improvements, and fixes across GitHub products, all in one place.
Businesses
By Stage
Enterprises
Team
Startups
Nonprofits
By Industry
Healthcare
Financial services
Manufacturing
Government
See all industries
Evaluate GitHub
Why GitHub
Customer Stories
Business insights
Trust Center
Add-ons
GitHub Advanced Security
Enterprise-grade code security
Copilot for business
Enterprise-grade AI features
Premium Support
Enterprise-grade 24/7 support
Resources
Learn & Connect
Documentation
Resource center
Blog
Events & webinars
Community Forum
Customer Stories
Integrations
Support
Customer support
Training
GitHub Status
Company
About
Partners
Changelog
GitHub Universe
GitHub Universe
The event for developers, by developers — talks, workshops, and what’s next for GitHub.
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 }}
mw00
project-maya
Repository navigation
Code
Issues
92
(92)
Pull requests
98
(98)
Discussions
Actions
Security and quality
Insights
More
items
All pull requests
Labels
20
(20)
Milestones
0
(0)
New pull request
Search pull requests
is
:
pr
state
:
open
is:pr state:open
Clear filter
Search
Pull requests
Open
98
(98)
Closed
506
(506)
Author
Label
Projects
Milestones
Reviews
Assignee
Sort by
Newest
descending
Sort…
Sort by Newest descending
More items
Comfortable display density
Compact display density
Hy4-Flash: add opt-in runtime refusal-direction projection
#1088
·
tctiger
opened
Nov 21, 2026
Contributor
·
1 task done
·
·
Hy4-Flash: add opt-in Engram table locking on Linux
#1087
·
tctiger
opened
Nov 21, 2026
Contributor
·
·
DeepSeek EXL3 on two tensor-parallel groups: decode 55 -> 70 tok/s after a model switch, long prompts as fast (DS4_SPLIT n@g, NUMA room at load)
#1083
·
needmorevram
opened
Nov 20, 2026
Contributor
·
·
hy4: eliminate CPU-lane barrier convoy and parallelize HY4_HOT gather
#1082
·
lastbl0c
opened
Nov 20, 2026
·
·
Inkling 2, phase 2: maya-ik2 - the engine, served by Maya's server (#974)
#1079
·
OscarSheng
opened
Nov 20, 2026
Contributor
·
·
ds4: DS4_AR_P2P on HIP with uncached buffers instead of the #809 lockout
#1072
·
paltian
opened
Nov 20, 2026
Contributor
·
4 of 5 tasks
·
·
DeepSeek: run resident EXL3 experts on the helper during prefill
#1065
·
LostVector
opened
Nov 20, 2026
Contributor
·
·
ref_gguf: --workers N, the low-bit routed experts in CPU worker processes
#1061
·
sramteare
opened
Nov 20, 2026
·
4 of 5 tasks
·
·
glm fast: WORKERS_FAST + default-off spec-pipe / MoE / pruned MTP-sample micro-optimizations
#1054
·
saatvik-viniak
opened
Nov 19, 2026
Contributor
·
·
1
GLM-5.3 (glm-dsa): KV streaming that fits the 744B's layer schedule. One residency map per indexer group, an identity window, prompts inside the window without staging
#1052
·
somnergy
opened
Nov 19, 2026
Contributor
·
·
Vision: the image encoder on a card of your choice ("vision": "gpu": …
#1050
·
acs48
opened
Nov 19, 2026
·
3 of 5 tasks
·
·
1
Add GGML Q5_1 (type 7) to the native MMVQ path and BF16 dequant
#1049
·
zehm
opened
Nov 19, 2026
Contributor
·
·
HIP: opt-in MFMA wave64 MLA prompt attention for CDNA (MAYA_GLM_PREFILL_ATTN=mfma)
#1048
·
zehm
opened
Nov 19, 2026
Contributor
·
·
DeepSeek on Windows: reuse RAM-store experts for recent-cache fills (opt-in)
#1046
·
haomin1996
opened
Nov 18, 2026
·
9 of 11 tasks
·
·
DeepSeek RAM store: slots of each expert's own size
#1045
·
woodsonl
opened
Nov 18, 2026
·
4 of 5 tasks
·
·
ds4: DS4_PROMPT_LIGHT_SMALL - a smaller light cut for short prompt chunks (opt-in, on #967)
#1034
·
pepuscz
opened
Nov 17, 2026
Contributor
·
·
ds4: DS4_SETTLE_POOL_PROMPT - prompt reads' picks steer the pool's idle leg (opt-in, on #1032)
#1033
·
pepuscz
opened
Nov 17, 2026
Contributor
·
·
1
ds4: DS4_SETTLE_POOL - the idle settle's third leg: the hottest file experts into the helper's pool (opt-in, on #1002)
#1032
·
pepuscz
opened
Nov 17, 2026
Contributor
·
·
1
ds4: DS4_LIGHT_X_ONCE - the light experts' input rows to the helper once a token (opt-in, same bits, on #1003)
#1031
·
pepuscz
opened
Nov 17, 2026
Contributor
·
·
ds4: DS4_EG_LATE - a one-chunk prompt's layers start before its Engram rows are all read (opt-in, same bits)
#1030
·
pepuscz
opened
Nov 17, 2026
Contributor
·
·
ds4: DS4_HC_MIX_ROWS - a prompt chunk's mHC mixes in one kernel instead of cuBLAS SGEMM (opt-in)
#1029
·
pepuscz
opened
Nov 17, 2026
Contributor
·
·
Users: accounts in SQLite, admin and standard roles, API keys of each…
#1028
·
drmatrix
opened
Nov 17, 2026
·
1 of 5 tasks
·
·
GLM: opt-in learned-prefix and completion-tail caching
#1025
·
oscarlius
opened
Nov 17, 2026
Contributor
·
3 of 6 tasks
·
·
5
Pinned host memory on explicit 2 MB huge pages, from a reserved or an on-demand pool
#1024
·
cresslank
opened
Nov 17, 2026
Contributor
·
·
7
GLM-5.3 (744B): KV streaming as opt-in on the Models page and by hand (the setup's default stays off)
#1023
·
BrandoB
opened
Nov 17, 2026
·
4 of 5 tasks
·
·
Previous
1
2
3
4
Next
You can’t perform that action at this time.