๊ฐ€์ž… ํ›„ ์ดˆ๋Œ€ ๋งํฌ๋ฅผ ๊ณต์œ ํ•˜๋ฉด ๋™์˜์ƒ ์žฌ์ƒ ๋ฐ ์ดˆ๋Œ€ ๋ณด์ƒ์„ ๋ฐ›์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

DoฤŸaรง
@dogacel0
ML Research @fal โ€ข MSc @NorthwesternU โ€ข Prev. SWE @CarbonHealth
๊ฐ€์ž… May 2019
622 ํŒ”๋กœ์ž‰ ์ค‘    1.1K ํŒฌ
Excited to announce our work "Attention Drift" has been accepted to NeurIPS 2026 ๐ŸŽ‰ I'm so happy to see our method's widespread adoptation. Now supported by all major inference engines, vLLM, Tokenspeed, SGLang and TRT-LLM. Also community trained speculative decoding models for Minimax M3 (@inferact and @AMD) and Kimi K2.6-K3 trained (@lightseekorg) being served in production as EAGLE 3.1 Shout out to all the great team mebers @themishtimoi @yigitcancomlek @kaan_oktay @hongyangzh and Stephen Xia. And special thanks to @fal for helping for making this research possible ๐Ÿ™‚
๋” ๋ณด๊ธฐ