Install
Please confirm you are human
This browser or connection looks automated. Press and continuously hold the control for 3 seconds to enable Google-hosted web results and, when separately allowed, AI-assisted answers.
A successful check enables 100 search requests. Interactive access does not authorize scraping, systematic collection, or reuse of search output.
News
When a Zero-Parameter Cache Overtakes a Transformer
30+ min ago (574+ words) The question this series exists to answer, put in a form that can be measured: a transformer sees a fixed 64-token window, a count table over the current document sees the whole document, and as documents get longer, how much…...
I Rewrote One Function in Squirrel, Then in C. One Line of Python Was Just as Fast.
4+ hour, 52+ min ago (543+ words) I maintain a 2D game engine in Python called ABS Engine. Last week I decided it needed C. Not because anything was slow. That is the embarrassing part. I wanted to drop a.c file into a folder and call it from Python…...
LLM Discussions (02): Train-time Vs. Test-time Compute
1+ hour, 32+ min ago (155+ words) Small Models with Deep Thinking Out-perform Frontier Models — But How? If this is the first story you have come across in our LLM Discussions …...
How Seed4J Validates Compatibility
7+ hour, 25+ min ago (279+ words) How Seed4J uses Renovate, GitHub Actions, and 25 generated project configurations to validate compatibility across modern technology stacks. Tagged with opensource, githubactions, devops, security....
BPE-Style Tokenizers: The Small Algorithm That Decides What an LLM Can See
8+ hour, 17+ min ago (1289+ words) Hello, I'm Shrijith Venkatramana, and I'm building LiveReview — a blast-radius aware AI code review... Tagged with ai, webdev, programming, productivity....
Ship fewer JS threads — a React Native interaction budget
8+ hour, 43+ min ago (170+ words) Users forgive a slow cold start more than a laggy tap. After the first screen, every interaction is a promise: press → feedback within ~100ms, content within a few hundred. On a mid-range Android (or an emulator profile that matches one), time…...
Our users’ phones were the backup: rebuilding a lost database from app caches
4+ hour, 54+ min ago (223+ words) How expired cache entries on readers’ devices, server logs, and a year-old dump were merged back into MongoDB without …...
4,768 LLM Runs, Zero Lost Sweeps: Hardening a Field-Test Runner for Timeouts, Hangs, and Cost
9+ hour, 22+ min ago (593+ words) CauterRule is an open-source sidecar that learns standing rules from repeated agent failures. It extracts lessons from trajectories, replay-tests them, and tries to separate reusable guidance from noisy overgeneralization. Here is the failure mode that never makes it into a…...
Remote Chat Failover Needs a Tab-Stop Origin Banner, Not a Skeleton Overlay
10+ hour, 23+ min ago (999+ words) I was halfway through a keyboard-only pass on a streaming chat when the local path died. A skeleton overlay covered the composer, my focus jumped, and tokens started arriving from somewhere else. Did the interface tell me those tokens had…...
Prompt Caching: Why cache_control Writes But Never Reads
10+ hour, 38+ min ago (457+ words) I turned on prompt caching for an agent loop that resends a 12K-token system prompt on every turn. Obvious win, right? Input tokens are the whole bill in a tool loop. The bill went up. Not a little. Roughly a…...