Skip to content

Pull requests: andrewkchan/deepseek.cpp

New pull request New
Clear current search query, filters, and sorts
1 Open 12 Closed
1 Open 12 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

add nucleus (top-p) sampling
#17 by andrewkchan was merged Jun 3, 2025 Loading…
add --bsize arg to convert.py to allow configuring f8e5m2 block size
#16 by andrewkchan was merged May 26, 2025 Loading…
Fix quality degradation after 4096 tokens
#15 by andrewkchan was merged May 26, 2025 Loading…
Test perplexity on wikitext-2
#14 by andrewkchan was merged May 31, 2025 Loading…
1
Refactor weights from raw pointers to tensor structs
#12 by andrewkchan was merged May 15, 2025 Loading…
add interactive mode
#11 by andrewkchan was merged May 10, 2025 Loading…
Add 3-bit K-quants
#10 by andrewkchan was merged May 6, 2025 Loading…
refactor Block into BlockMHA, BlockMLA
#9 by andrewkchan was merged May 3, 2025 Loading…
Add multi-latent attention, profiling instrumentation, other perf fixes
#8 by andrewkchan was merged May 2, 2025 Loading…
Add llama.cpp Q2_K support
#7 by andrewkchan was merged Apr 20, 2025 Loading…
optimize blockwise scale reads in matmuls
#3 by andrewkchan was merged Feb 11, 2025 Loading…
Add deepseek v3 support and blockwise quantization
#1 by andrewkchan was merged Feb 6, 2025 Loading…
ProTip! Exclude everything labeled bug with -label:bug.