Skip to content

Pull requests: lightseekorg/tokenspeed

New pull request New
34 Open 278 Closed
34 Open 278 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 Qwen3.5 MoE text config initialization
#328 opened Jun 1, 2026 by lightseek-bot Contributor Loading…
6
feat(engine): compute_log_probs API for RL sequence scoring (RL-plan M2)
#321 opened May 30, 2026 by HJSang Collaborator Loading…
9
fix(eagle): avoid mutating drafter sequence lengths
#318 opened May 30, 2026 by XucSh Contributor • Draft
6
[WIP]perf: add Gluon MoE kernels for GPT-OSS
#314 opened May 29, 2026 by knwng Contributor Loading…
9
[Perf] Optimizes loads in gfx950 fp16 decode kernel
#313 opened May 29, 2026 by Yu-Zhewen Loading…
3
test: extend agentic bench
#312 opened May 29, 2026 by syuoni Member Loading…
fix(cute_dsl): skip CuTe DSL argmax kernel on H20 GPUs
#311 opened May 29, 2026 by botieking98 Loading…
1
3
fix(dp): fix qwen 3.5 data parallel bug.
#309 opened May 29, 2026 by tuanzhangCS Contributor Loading…
2
feat(entrypoints): add HTTP server sidecar alongside smg gateway
#308 opened May 29, 2026 by qywu Collaborator Loading…
8
[WIP] refactor + perf(spec-decode): refactor #217 + add prefill scope
#304 opened May 28, 2026 by rjzhb Contributor Loading…
23
(feat) L3 KVStore: prefetch and backup support
#293 opened May 28, 2026 by ehuohz Loading…
@XucSh
10
perf: Optimize inter-iteration small op
#291 opened May 28, 2026 by yweng0828 Contributor Loading…
4
[Perf] Enable missing FlashInfer MoE autotuning
#290 opened May 28, 2026 by mmangkad Contributor Loading…
8
perf(model_loader): multi-threaded safetensors weight loading
#287 opened May 28, 2026 by yuanqingz Loading…
6
fix(PD): fix PD speculative bootstrap input seeding
#286 opened May 28, 2026 by XucSh Contributor Loading…
8
chore: use -O3 -use_fast_math for tokenspeed_kernel compilation
#285 opened May 28, 2026 by syuoni Member Loading…
2
Add Triton sampling backends alongside FlashInfer
#280 opened May 27, 2026 by FlamingoPg Contributor Loading…
17
feat(memory-saver): optional CPU staging for round-trip weight preservation
#275 opened May 27, 2026 by qywu Collaborator • Draft
3 of 5 tasks
feat(memory-saver): wrap CUDA graphs and attention workspaces in saver.region()
#274 opened May 27, 2026 by qywu Collaborator • Draft
3 of 5 tasks
feat(scheduler): per-adapter KV prefix-cache namespace + max_loras batch cap
#268 opened May 26, 2026 by qywu Collaborator Loading…
5 tasks
38
feat(spec-decode): add native DFlash support
#263 opened May 26, 2026 by mesaleh Loading…
23
fix(deepseek): guard missing quant weight_block_size
#261 opened May 26, 2026 by mesaleh Loading…
fix(trtllm-mla): make spec-decode CUDA graph capture causal
#260 opened May 26, 2026 by mesaleh Loading…
6
perf: TokenSpeed MLA decode kernel optimization for num_heads=16
#255 opened May 26, 2026 by dishengbin Contributor • Draft
5
feat(eplb): eplb support high priority
#251 opened May 26, 2026 by XucSh Contributor Loading…
11
ProTip! Filter pull requests by the default branch with base:main.