All news

RemoteANN does nearest-neighbor search directly on object storage with up to 16.1x fewer requests and 5.9x higher throughput than baselines.

Tests whether self-evolving agents like ReasoningBank, GEPA and Meta-Harness learn new tasks without forgetting old ones.

Evolving the interfaces that computer-use agents see makes them up to 3.8x better.

FlashMap uses flow maps to improve the speed and acceptance trade-off in speculative decoding.

Finds numerical bugs in standard softmax and cross-entropy implementations, like softmax collapse, that change training dynamics, and proposes fixes.

Kronecker Delta Rule treats key-value pairs as joint addresses in linear attention to reduce interference at long context.

Coding agents follow stale conventions in instruction files over the actual code. A one-sentence prompt fix telling them to prefer existing code improves decisions a lot.

SuperMimic uses one differentiable world model for humanoid motion control.

Tests whether frontier models can autonomously build and improve their own software harness.

DiscoveryRSI uses recursive learning to find testable scientific hypotheses.

Evaluates the non-generative Jev model as a relevance judge and reranker for Portuguese IR.