Tag archive
A new agent harness lifts Qwen 3.7-Plus from 51.8% to 80.7% on a long-horizon coding benchmark without touching the model, by keeping task state outside the conversation and updating it only from facts a read-only auditor verified in the environment.
Eviction and summarization keep a long-running agent's window small, but the summaries themselves still accumulate in the linear transcript — and on a run long enough, that accumulation becomes the new quadratic. Past a few hundred steps, the fix isn't a better compaction policy inside the window. It's moving history out of the window entirely and querying it like a database instead of replaying it like a log.
A naive agent's token bill doesn't grow with the number of steps — it grows with the square of them, because every step re-reads the whole transcript that every previous step appended to. A small cost model shows the curve, and four structural moves turn the quadratic back into something close to linear without dropping information the agent actually needs.
Summary What is it: Imagine you have an agent that has a lot of different jobs to do...
A new framework called LaMem-VLA tackles the 'goldfish memory' problem in robot policies by compressing past experience into latent memory tokens and weaving them into the robot's current reasoning, targeting long-horizon manipulation tasks that sing
New research reframes agent memory as a trainable skill rather than a growing transcript - and shows that optimizing memory alone lets a 32B open model rival Claude Opus, and a 4B model leap from 4% to 78% on a benchmark.
Why Long-Horizon Existence Requires an Ontology Dissecting Navigational...