Skip to content

Pull requests: lightseekorg/tokenspeed

New pull request New
29 Open 523 Closed
29 Open 523 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

fix(attention): treat window_left as exclusive
#578 opened Jul 1, 2026 by Yu-Zhewen Contributor Loading…
2
Security: Potential race condition in global backend initialization
#575 opened Jul 1, 2026 by tomaioo Loading…
2
Fix DP cache sizing
#574 opened Jul 1, 2026 by tuanzhangCS Contributor Loading…
perf(deepseek-v4): accelerate indexer Q and packed FP8 quantization
#563 opened Jun 30, 2026 by dongjiyingdjy Contributor Loading…
perf(multimodal): Prewarm Kimi vision encoder before readiness
#556 opened Jun 30, 2026 by qimcis Contributor Loading…
1
perf(scheduler): Split mixed prefill/decode forwards to preserve decode graphs
#555 opened Jun 30, 2026 by qimcis Contributor Loading…
3
feat(serve): export Prometheus metrics via /metrics endpoint
#553 opened Jun 29, 2026 by ilyaters Loading…
4
feat(platform): model the Blackwell family — datacenter / Thor / consumer
#551 opened Jun 28, 2026 by jasl Contributor Loading…
1
docs: KV events + Mooncake Store integration plan
#550 opened Jun 28, 2026 by stmatengss • Draft
ci(disaggregation): add Qwen3.5 EPD encode coverage
#549 opened Jun 28, 2026 by chenht2022 Contributor • Draft
feat(disaggregation): add EPD encode pipeline
#548 opened Jun 28, 2026 by chenht2022 Contributor Loading…
11
perf(deepseek-v4): integrate TRT-LLM mHC kernels
#547 opened Jun 28, 2026 by dongjiyingdjy Contributor • Draft
feat(rl): weight-transfer control plane for RL online weight sync
#546 opened Jun 28, 2026 by qywu Collaborator Loading…
15
refactor(spec-decode): simplify deepseekV3/GLM attention path for #217 (3/3)
#544 opened Jun 28, 2026 by rjzhb Collaborator Loading…
2
test: glm-5.2 agentic bench
#532 opened Jun 26, 2026 by syuoni Member • Draft
feat: support qwen3.5 on Hopper GPUs
#520 opened Jun 25, 2026 by XuZhang99 • Draft
[WIP] feat:support qwen3.5 dflash
#510 opened Jun 24, 2026 by minedec Contributor • Draft
3
test(agentic): add EvalScope trie benchmark protocol
#466 opened Jun 17, 2026 by Xiangyi1996 Collaborator • Draft
test(ci): add DeepSeek-V4-Flash MTP AIME25 eval
#461 opened Jun 16, 2026 by dongjiyingdjy Contributor Loading…
2
fix(scheduler): release paged-cache snapshots in ~HybridPrefixCache to avoid teardown use-after-free inactive
#455 opened Jun 15, 2026 by Sunt-ing Loading…
1
test: add dp4ep4 case in CI
#453 opened Jun 15, 2026 by tuanzhangCS Contributor Loading…
4
[WIP] Refactor Cache Management
#447 opened Jun 15, 2026 by wangbo981016 Contributor • Draft
Fix EP8 DP/TP RSAG init and empty LM head
#416 opened Jun 11, 2026 by yubofredwang Contributor Loading…
2
perf(gdn): fuse causal_conv1d and QKV split for GDN prefill
#382 opened Jun 8, 2026 by elwhyjay Contributor Loading…
10
fix(cache): Coarsely fence the compute stream behind the host loadback stream on.
#370 opened Jun 6, 2026 by LorrinWWW Contributor Loading…
4
ProTip! What’s not been updated in a month: updated:<2026-06-01.