-
Notifications
You must be signed in to change notification settings - Fork 251
Pull requests: lightseekorg/tokenspeed
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
perf(attn_res): remove Blackwell 16K token limit
#1219
opened Aug 23, 2026 by
chenht2022
Contributor
Loading…
perf(metadata): scrub the padded input tail in one launch
#1218
opened Aug 23, 2026 by
nperrin-fr
Collaborator
•
Draft
test(kimi-k3): PD-disaggregation simulation bench (pd_sim)
#1216
opened Aug 23, 2026 by
dongjiyingdjy
Contributor
•
Draft
perf(amd): move gfx1250 KDA prefill/decode to V-major
#1215
opened Aug 23, 2026 by
Yu-Zhewen
Contributor
Loading…
[WIP] feat: add DFlash2 draft serving support
#1211
opened Aug 23, 2026 by
torchspec-bot
Collaborator
•
Draft
feat(dspark): expose forced-rejection diagnostics
#1170
opened Aug 20, 2026 by
jhinpan
Contributor
Loading…
fix(cache): route null decode pages to dummy slot
#1169
opened Aug 20, 2026 by
jhinpan
Contributor
Loading…
fix(cache): support attention-DP for Kimi-K3 by deriving the MLA packing from the KDA state size
#1152
opened Aug 19, 2026 by
dongjiyingdjy
Contributor
Loading…
(WIP) perf(kimi3): tune small-batch latent projection
#1135
opened Aug 19, 2026 by
panditsa
Contributor
Loading…
feat(cache): add Mooncake Store as L3 under compact Host KV
#1125
opened Aug 17, 2026 by
stmatengss
Loading…
feat(engine): surface is_sleeping as a GET control endpoint mirroring /is_paused
#1118
opened Aug 16, 2026 by
SuperMarioYL
Loading…
feat(zmq): materialize precomputed multimodal inputs on msgpack ingest
#1081
opened Aug 13, 2026 by
slin1237
Contributor
Loading…
Previous Next
ProTip!
Follow long discussions with comments:>50.