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 }}
ashhart
TensorFold
Repository navigation
Code
Issues
192
(192)
Pull requests
395
(395)
Actions
Projects
Security and quality
Insights
More
items
All pull requests
Labels
11
(11)
Milestones
0
(0)
New pull request
Search pull requests
is
:
pr
state
:
open
is:pr state:open
Clear filter
Search
Pull requests
Open
395
(395)
Closed
3,246
(3,246)
Author
Label
Projects
Milestones
Reviews
Assignee
Sort by
Newest
descending
Sort…
Sort by Newest descending
More items
Comfortable display density
Compact display density
train 1122171958: #4655 (CI only)
#4659
·
ashhart
opened
Nov 22, 2026
Owner
·
·
train 1122170937: #4653 #4654 (CI only)
#4658
·
ashhart
opened
Nov 22, 2026
Owner
·
·
train 1122162855: #4606 #4617 #4628 #4629 (CI only)
#4657
·
ashhart
opened
Nov 22, 2026
Owner
·
·
train 1122162916: #4648 (CI only)
#4656
·
ashhart
opened
Nov 22, 2026
Owner
·
·
tests: DeepSeek-V4.1 chat and fixture tests pin the settings they assume
#4655
·
ashhart
opened
Nov 22, 2026
Owner
·
·
tests: the admission stress test waits for its choices to submit before a client walks away
#4654
·
ashhart
opened
Nov 22, 2026
Owner
·
·
sessions: the tier thread's job runner can't lose a wakeup (the DeepSeek sessions test's 20-minute hang)
#4653
·
ashhart
opened
Nov 22, 2026
Owner
·
·
XPU Qwen3.8-27B: decode attention takes a key's twelve exponentials as one vector exp, bit-equal
#4651
·
Mirrdhyn
opened
Nov 22, 2026
Contributor
·
·
Upstream PR: long image renders give way to live speech on the GPU
#4650
·
plotarmordev
opened
Nov 22, 2026
Contributor
·
2 of 4 tasks
·
·
Inkling 2 Mini on Metal: wide prompt chunks take the routed experts on turned tiles, 16 weight rows by 32 of an expert's pairs an op
#4649
·
sethforprivacy
opened
Nov 22, 2026
Contributor
·
8 tasks done
·
·
qwen3.5 cuda dflash2: opt-in batched drafter absorb across a round's streams
#4648
·
ashhart
opened
Nov 22, 2026
Owner
·
·
train 1122153311: #4617 (CI only)
#4646
·
ashhart
opened
Nov 22, 2026
Owner
·
·
Inkling 2 Mini on Metal: the prompt kernel reads its key tiles from the cache on M5 (a round at 96K context 28.6 to 21.5 ms, a 45K prompt 1,892 to 2,078 tok/s)
#4645
·
sethforprivacy
opened
Nov 22, 2026
Contributor
·
8 tasks done
·
·
Inkling 2 Mini on Metal: prompt chunks of up to 1,024 rows, their taps staged in the scratch (a 45K prompt 1,634 to 1,892 tok/s)
#4644
·
sethforprivacy
opened
Nov 22, 2026
Contributor
·
8 tasks done
·
·
train 1122150250: #4642 (CI only)
#4643
·
ashhart
opened
Nov 22, 2026
Owner
·
·
Upstream PR: H3 Metal products on h3_gemm_32x64 in 64-row bands, 64-tile decoder pass
#4641
·
plotarmordev
opened
Nov 22, 2026
Contributor
·
5 tasks done
·
·
Inkling 2 Mini on CUDA: #4387's engine loads the checkpoint, answers correctly (64/64 teacher-forced), serves --parallel streams, and resumes prompt-cache states exactly
#4640
·
jschmied
opened
Nov 22, 2026
Contributor
·
7 of 8 tasks
·
·
Inkling 2 Mini on Metal: a window's grammar masks step by the padded vocabulary, so a structured reply passes its second row
#4635
·
sethforprivacy
opened
Nov 22, 2026
Contributor
·
8 tasks done
·
·
qwen27 metal: serve the tied 0.8B and 4B, as CUDA does
#4634
·
joshuaswarren
opened
Nov 22, 2026
Contributor
·
·
1
feat(hip): Inkling 2 attention kernels under the kernel contract
#4633
·
AbrahamCete
opened
Nov 22, 2026
Contributor
·
4 of 8 tasks
·
·
flashnext cuda: the MTP rows' row best runs over 32 blocks a row, fn_row_best's bits
#4632
·
jayleaton
opened
Nov 22, 2026
Contributor
·
6 of 8 tasks
·
·
Inkling 2 Mini on Metal: a prompt's projections and sliding attention on the tensor units (a 45K prompt 906 to 1,634 tok/s)
#4631
·
sethforprivacy
opened
Nov 22, 2026
Contributor
·
8 tasks done
·
·
qwen27 metal dflash2: opt-in batched drafter absorb across streams
#4629
·
ashhart
opened
Nov 22, 2026
Owner
·
·
qwen27 metal dflash2: opt-in adaptive drafter block for a lone stream
#4628
·
ashhart
opened
Nov 22, 2026
Owner
·
·
gemma4 metal: opt-in fused MLP for drafted rounds of 2-16 rows
#4624
·
ashhart
opened
Nov 22, 2026
Owner
·
·
Previous
1
2
3
4
5
6
7
…
16
Next
You can’t perform that action at this time.