LoreonLabsPlatform

Overview

Intelligence

Markets
Builders
Research
Narratives
Ecosystems
Launchpads

Discover

Search
Sources

Other

tokenspeed

TokenSpeed is a speed-of-light LLM inference engine.

OtherEmerging

Stars

—

Forks

—

Contributors

8

Last push

14d ago

Recent commits

Latest commits.

feat: reduce DeepSeek V4 prefix state snapshots with replay reuse (#329)
1176731Simon_CQK14d ago
perf: add Gluon MoE kernels for GPT-OSS (#314)
1780377Kyle Wang15d ago
perf(deepseek-v4): decode attention optimizations (#339)
b80268fdongjiyingdjy15d ago
Defer perf reference failure reporting (#340)
e26e7a1Yineng Zhang15d ago
perf(kernel): optimize mha kernel for sliding window case (#336)
94ccce6Pengzhan Zhao15d ago

Fix(spec decode): catch up trim bug (#335)

15385b2Hongbin Zhong15d ago

fix(PD): fix PD speculative bootstrap input seeding (#286)

58a5993Xuchun Shang15d ago

chore: use -O3 -use_fast_math for tokenspeed_kernel compilation (#285)

1e2ab88Enwei Zhu16d ago

Top contributors

Builders behind this project.