Cartridges: 'Self-Study is a Moving Target' — a label-free metric for KV-cache compression
simran_s_arora · x · 2026-11-05
Ramp Labs published 'Self-Study is a Moving Target' by researcher Ben Geist.
Cartridges compress a long document into a small trained KV cache via self-study — training on a series of self-generated questions about the document. The article derives a label-free measure of question quality, arguing that self-study is a moving target: the question distribution shifts as training progresses, so one must dynamically decide which questions are worth training on. Worth reading in full for anyone working on long-context or memory compression.
More from Research
- GitHub Next's PDD Kit uses formal methods to prove AI-generated web app code is correct — 0xkarasy · 2026-11-05
- Meta's JiTTesting deploys LLM-generated tests to catch bugs pre-merge, cutting human review 70% — AxSaucedo · 2026-11-05
- 演示显示 LLM 能在普通散文里藏密码,人类肉眼难察觉 — dosco · 2026-11-05
- Robot benchmarks slammed as fantasy: real robots can't afford $10k GPUs for VLA models — AIFlow_ML · 2026-11-05
- Agents ace isolated tasks but drop to ~38% at maintaining codebases over time — FullStackFrenzy · 2026-11-05
- RoboColosseum founder: robotics lags AI because it lacks real-world benchmarks — DJiafei · 2026-11-05