kkm's submissions | Hacker News

1.		Aleph Alpha's latest Model playing Doom (tesseracted.com)
		1 point by kkm 25 days ago \| past \| 1 comment
2.		Aleph Alpha's Kolibri-1 playing Breakout – Making decisions under 25ms per move (twitter.com/konarkmodi)
		1 point by kkm 29 days ago \| past
3.		How to serve trillions of tokens for trillion-parameter coding agents (modal.com)
		1 point by kkm 42 days ago \| past
4.		Unlocking parallel test-time scaling for long-horizon agents (doubleword.ai)
		3 points by kkm 43 days ago \| past
5.		Claude.md is good for taste and project context.It's a weak place for invariants (tesseracted-labs-blog.vercel.app)
		3 points by kkm 49 days ago \| past
6.		Moving coding-agent guardrails from prompts to hooks (tesseracted-labs-blog.vercel.app)
		4 points by kkm 50 days ago \| past
7.		Improving Throughput by Optimising KV Cache Efficiency for Agentic Workloads (j9s.io)
		3 points by kkm 51 days ago \| past \| 1 comment
8.		Guide to the Kimi DeltaNet Family of linear attention (doubleword.ai)
		3 points by kkm 3 months ago \| past
9.		Forensic Analysis of Container Snapshot Chains for Post-Event Reconstruction [pdf] (radostin.io)
		2 points by kkm 3 months ago \| past
10.		The Agent swarm that designs itself (peterbhabra.com)
		1 point by kkm 4 months ago \| past
11.		Don't Build a Router. Train the Small Model to Know When to Defer (distillabs.ai)
		2 points by kkm 4 months ago \| past \| 1 comment
12.		The gap between open weights LLMs and closed source LLMs (doubleword.ai)
		306 points by kkm 4 months ago \| past \| 250 comments
13.		InfiniBand, RoCE, and All That (fergusfinn.com)
		5 points by kkm 4 months ago \| past
14.		2678x Faster Matrix Multiplication with a GPU (0mean1sigma.com)
		2 points by kkm 4 months ago \| past
15.		UCCL-EP: DeepEP-style expert parallelism on any NIC, no GPU-initiated comms (fergusfinn.com)
		9 points by kkm 4 months ago \| past
16.		Hacking Google with A.I. For $500k (brutecat.com)
		1 point by kkm 4 months ago \| past
17.		How to setup a local coding agent on macOS (ikyle.me)
		507 points by kkm 4 months ago \| past \| 127 comments
18.		Anatomy of a high-performance EP kernel (fergusfinn.com)
		16 points by kkm 4 months ago \| past \| 1 comment
19.		No Token Left Behind: Demystifying Token-in-Token-Out in Miles (lmsys.org)
		2 points by kkm 4 months ago \| past
20.		MoE expert co-activations: Reordering inputs yields easy throughput gains (doubleword.ai)
		2 points by kkm 4 months ago \| past
21.		The Economics of Speculative Decoding (fergusfinn.com)
		30 points by kkm 4 months ago \| past \| 6 comments
22.		Speculative KV coding: losslessly compressing KV cache by up to ~4× (fergusfinn.com)
		155 points by kkm 5 months ago \| past \| 48 comments
23.		70x faster cold(ish) starts for SGLang (fergusfinn.com)
		1 point by kkm 5 months ago \| past
24.		Bringing Up DeepSeek-V4-Flash on AMD MI300X (fergusfinn.com)
		120 points by kkm 5 months ago \| past \| 25 comments
25.		Brave AI privacy:LLMs on NEAR AI Nvidia-Backed Trusted Execution Environments (brave.com)
		1 point by kkm 11 months ago \| past
26.		How fast can an LLM go? (fergusfinn.com)
		2 points by kkm 11 months ago \| past
27.		FHE can be leveraged for LLMs such as ChatGPT in a privacy-preserving manner (huggingface.co)
		4 points by kkm on Aug 13, 2024 \| past
28.		Harnessing the Power of Large Language Models for Insightful Review Analysis (holidaycheck.com)
		1 point by kkm on April 16, 2024 \| past
29.		A Privacy-First approach to use AI for understanding our Customers Better (holidaycheck.com)
		1 point by kkm on March 12, 2024 \| past
30.		How to make LLMs go fast (vgel.me)
		2 points by kkm on Dec 19, 2023 \| past
		More