HackerTrans
TopNewTrendsCommentsPastAskShowJobs

somnial

7 karmajoined 9개월 전

Submissions

The Anatomy of an Instruction Pipeline Hazard

hiraditya.github.io
5 points·by somnial·4시간 전·0 comments

Width vs. Depth: Speculating on the Margin

blog.doubleword.ai
17 points·by somnial·12일 전·1 comments

Pushing memory bound CUDA kernels past the speed of light with data compression

fergusfinn.com
2 points·by somnial·2개월 전·0 comments

Speculative KV coding: ~4× losslessly compressed KV cache using a small model

fergusfinn.com
2 points·by somnial·2개월 전·0 comments

[untitled]

1 points·by somnial·2개월 전·0 comments

70x faster cold(ish) starts for SGLang

fergusfinn.com
1 points·by somnial·3개월 전·0 comments

LLM powered data structures: A lock-free binary search tree

fergusfinn.com
1 points·by somnial·6개월 전·0 comments

Parallel Primitives for Multi-Agent Workflows

fergusfinn.com
1 points·by somnial·6개월 전·0 comments

Scheduling in LLM Inference

fergusfinn.com
1 points·by somnial·8개월 전·0 comments

How fast can an LLM go?

fergusfinn.com
2 points·by somnial·9개월 전·0 comments

comments

somnial
·17일 전·discuss
this is a blog post from a company that hosts open weights LLMs (https://www.doubleword.ai/). I think its possible it might have been tongue in cheek
somnial
·지난달·discuss
https://fergusfinn.com/blog/economics-of-speculative-decodin...

good point tho - plus for Deepseek the shared expert increases the overlap slightly
somnial
·지난달·discuss
true, but no reason the predictor model couldn't use linear attention (i.e. mamba, GDN etc) to predict KV caches