Research papers, read to you in five minutes.
Research papers, read to you in five minutes.

Agent trajectories collected under one harness poison SFT when you deploy under another—controller prompts and reflection scaffolding become crutches the model won't have. RSR has the base model rewrite multi-harness successes into clean runbook-guided re-executions under the target harness, lifting pass@3 ~20pp over direct SFT.

Agents · Added Oct 5

Predictive Credit: Measuring What Scientific Explanations Add to Experimental Forecasts

Do agent-written experiment rationales actually predict outcomes, or just sound good? A paired forecaster test across 336 states found matched explanations gave no confirmed accuracy gain over description-only or donor rationales, though structured prediction cards cut call-to-call drift ~60% — self-consistency without insight.

Agents · Added Oct 4

GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis

GraphForge tackles the real-files-vs-reliable-grading tradeoff in agent training data: it crawls actual documents, builds a cross-file evidence graph, then compiles rubrics whose criteria point back to specific files. Fine-tuning Qwen3 27B on just 2,169 trajectories lifts GDPVal by 65.7 Elo across three unseen scaffolds.

Diffusion · Added Oct 3

Adaptive Reward Routing: Dynamic Multi-Reward Optimization for Joint Audio-Video Diffusion via Forward-Process RL

Multi-reward RL on joint audio-video diffusion models silently kills weak objectives like A/V sync when gradient-based weighters zero them out, and static modality routing goes stale as cross-modal layers shift during fine-tuning. Adaptive Reward Routing re-routes gradients each round and floors user priors, cutting DeSync from 0.604 to 0.341.

Agents · Added Oct 3

Beyond Memory: Harnessing Long-Horizon Agents with Explicit Belief States

PoS tackles long-horizon agents acting on stale or self-contradictory world state by maintaining an explicit entity-state-relation belief, auditing updates, and detecting "belief trapping" via a sliding-window health score. Wins all 12 benchmark cells with up to 37.89% gains on root-cause analysis, but costs 5× the tokens.

Never fall behind the literature again.

Free account. Follow topics, build your queue.

Sign up free

Agents · Added Oct 6

Scaling Trajectories for Complex Tasks through Recursive Self-Rewrite

Agent trajectories collected under one harness poison SFT when you deploy under another—controller prompts and reflection scaffolding become crutches the model won't have. RSR has the base model rewrite multi-harness successes into clean runbook-guided re-executions under the target harness, lifting pass@3 ~20pp over direct SFT.

Agents · Added Oct 5

Predictive Credit: Measuring What Scientific Explanations Add to Experimental Forecasts

Do agent-written experiment rationales actually predict outcomes, or just sound good? A paired forecaster test across 336 states found matched explanations gave no confirmed accuracy gain over description-only or donor rationales, though structured prediction cards cut call-to-call drift ~60% — self-consistency without insight.

Agents · Added Oct 4

GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis

GraphForge tackles the real-files-vs-reliable-grading tradeoff in agent training data: it crawls actual documents, builds a cross-file evidence graph, then compiles rubrics whose criteria point back to specific files. Fine-tuning Qwen3 27B on just 2,169 trajectories lifts GDPVal by 65.7 Elo across three unseen scaffolds.

Diffusion · Added Oct 3

Adaptive Reward Routing: Dynamic Multi-Reward Optimization for Joint Audio-Video Diffusion via Forward-Process RL

Multi-reward RL on joint audio-video diffusion models silently kills weak objectives like A/V sync when gradient-based weighters zero them out, and static modality routing goes stale as cross-modal layers shift during fine-tuning. Adaptive Reward Routing re-routes gradients each round and floors user priors, cutting DeSync from 0.604 to 0.341.

Agents · Added Oct 3

Beyond Memory: Harnessing Long-Horizon Agents with Explicit Belief States

PoS tackles long-horizon agents acting on stale or self-contradictory world state by maintaining an explicit entity-state-relation belief, auditing updates, and detecting "belief trapping" via a sliding-window health score. Wins all 12 benchmark cells with up to 37.89% gains on root-cause analysis, but costs 5× the tokens.

Never fall behind the literature again.

Free account. Follow topics, build your queue.

Browse topics

Research papers, read to you in five minutes.Research papers, read to you in five minutes.

Research papers, read to you in five minutes.
Research papers, read to you in five minutes.