HackerTrans
TopNewTrendsCommentsPastAskShowJobs

rawsh

34 karmajoined قبل 6 سنوات

Submissions

Flash-MSA: Accelerating Million-Token Training with Sparse Attention Kernels

nanduruganesh.github.io
11 points·by rawsh·قبل 3 ساعات·0 comments

Zyphra ZAYA1-base: First large-scale model trained on AMD

zyphra.com
6 points·by rawsh·قبل 8 أشهر·1 comments

Debugging divergence between engine and transformers logprobs for RL

gist.github.com
2 points·by rawsh·قبل 10 أشهر·0 comments