HackerTrans
TopNewTrendsCommentsPastAskShowJobs

kkm

no profile record

Submissions

The Agent swarm that designs itself

peterbhabra.com
1 points·by kkm·11 दिन पहले·0 comments

Don't Build a Router. Train the Small Model to Know When to Defer

distillabs.ai
2 points·by kkm·11 दिन पहले·1 comments

The gap between open weights LLMs and closed source LLMs

blog.doubleword.ai
306 points·by kkm·15 दिन पहले·250 comments

[untitled]

1 points·by kkm·19 दिन पहले·0 comments

InfiniBand, RoCE, and All That

fergusfinn.com
5 points·by kkm·22 दिन पहले·0 comments

2678x Faster Matrix Multiplication with a GPU

0mean1sigma.com
2 points·by kkm·25 दिन पहले·0 comments

UCCL-EP: DeepEP-style expert parallelism on any NIC, no GPU-initiated comms

fergusfinn.com
9 points·by kkm·26 दिन पहले·0 comments

Hacking Google with A.I. For $500k

brutecat.com
1 points·by kkm·29 दिन पहले·0 comments

How to setup a local coding agent on macOS

ikyle.me
507 points·by kkm·29 दिन पहले·127 comments

Anatomy of a high-performance EP kernel

fergusfinn.com
16 points·by kkm·पिछला माह·1 comments

No Token Left Behind: Demystifying Token-in-Token-Out in Miles

lmsys.org
2 points·by kkm·पिछला माह·0 comments

MoE expert co-activations: Reordering inputs yields easy throughput gains

blog.doubleword.ai
2 points·by kkm·पिछला माह·0 comments

The Economics of Speculative Decoding

fergusfinn.com
30 points·by kkm·पिछला माह·6 comments

Speculative KV coding: losslessly compressing KV cache by up to ~4×

fergusfinn.com
155 points·by kkm·पिछला माह·48 comments

70x faster cold(ish) starts for SGLang

fergusfinn.com
1 points·by kkm·पिछला माह·0 comments

Bringing Up DeepSeek-V4-Flash on AMD MI300X

fergusfinn.com
120 points·by kkm·पिछला माह·25 comments

Brave AI privacy:LLMs on NEAR AI Nvidia-Backed Trusted Execution Environments

brave.com
1 points·by kkm·8 माह पहले·0 comments

How fast can an LLM go?

fergusfinn.com
2 points·by kkm·8 माह पहले·0 comments

comments

kkm
·पिछला माह·discuss
This is very interesting, planning to write about it?
kkm
·पिछला माह·discuss
Also the vllm patch accompanying the blogpost: https://github.com/doublewordai/vllm-amd-blog-doubleword