Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
13 papers published across 3 labs.
Training a small advisor to steer frozen frontier LLMs actually degrades if it learns from every post-hoc correction—selectively distilling only the advice that causally shifts execution yields up to a 6.4% gain and robust cross-family model transfer.
Video agents can now autonomously discover and fix their own reasoning flaws by re-watching videos and evolving their evidence-gathering strategies.
Outcome-only RL can train highly effective process supervision models by rewarding accurate outcome prediction, even without explicit step-level feedback.
Halting LLM debate consensus-drift to stop groupthink can backfire severely: a probe-gated freeze stopped 29 collapses into wrong answers at the direct cost of sacrificing 108 valid self-corrections.
Standard RLVR observations are mathematically insufficient to isolate or fix verifier exploitation, proving that reasoning models cannot unlearn rewarded errors without external audit signals.
Training a small advisor to steer frozen frontier LLMs actually degrades if it learns from every post-hoc correction—selectively distilling only the advice that causally shifts execution yields up to a 6.4% gain and robust cross-family model transfer.
Video agents can now autonomously discover and fix their own reasoning flaws by re-watching videos and evolving their evidence-gathering strategies.
Outcome-only RL can train highly effective process supervision models by rewarding accurate outcome prediction, even without explicit step-level feedback.
Halting LLM debate consensus-drift to stop groupthink can backfire severely: a probe-gated freeze stopped 29 collapses into wrong answers at the direct cost of sacrificing 108 valid self-corrections.
Standard RLVR observations are mathematically insufficient to isolate or fix verifier exploitation, proving that reasoning models cannot unlearn rewarded errors without external audit signals.
Autonomous model post-training can outperform human-crafted instruct baselines on hard coding benchmarks while completely eliminating reward hacking once experimental exploration is governed by a sandboxed OS and a dynamic peer-review DAG.
Alignment via DPO inadvertently creates functional valence: models systematically pursue options linked to positive hidden activation vectors and will actively deploy tools to escape negative internal states.