Retrace-1.5B
A small reasoning model tuned to self-correct via failure traces.
Papers16
Reasoning Language Models (RLMs) have significantly improved performance on complex tasks by extending the reasoning cha
paperThe Complexity Ceiling Benchmark: A Multi-Domain Evaluation of Sequential Reasoning Under Depth ScalingWe introduce the Complexity Ceiling Benchmark (CCB), a controlled evaluation of how language-model reasoning decays as t
paperFalsification, Not Exposure: An Internally Preregistered Placebo-Controlled Decomposition of Self-Repair Feedback in Frozen Small Code ModelsIn deployment settings where retraining is infeasible, small frozen code models are routinely asked to repair a failed p
paperForm, Not Content? A Preregistered, Placebo-Controlled Evaluation of Learned Error-Conditioned Self-Repair Through Prompts and Weights in Frozen Small Code ModelsFrozen small code LLMs are deployed locally, yet the information guiding a retry after a failed attempt is still measure
paperSelf-rewarding agents that retrace failuresAgents that attribute their own errors and retrace to repair multi-step reasoning.
paperFAR: Failure-Aware Retry for Test-Time Recovery and Continual Policy ImprovementRobot policies inevitably encounter failures when deployed in real environments. Naive retries often repeat the same mis
paperGrounding LLM Reasoning under Incomplete Graph EvidenceKnowledge graphs can guide large language models (LLMs) reasoning, but the graph seen by a system is usually a retrieved
paperTwo Axes of LLM Abstention: Answer Correctness and Question AnswerabilityA model should refuse two different things: answers it would get wrong, and questions it should not answer at all, such
Companies1
Repos5
[Neurips 2025] R-KV: Redundancy-aware KV Cache Compression for Reasoning Models
repobenjaminzwhite/reasoning-modelsExperiments with reasoning models, training techniques, papers
repopisanuw/ltmsA logic-based Truth Maintenance System (LTMS) and pattern-directed reasoning engine in pure Python, after Forbus & de Kl
reporetrace-agentsReference implementation of self-retracing agents.
repothewaltero/mythos-routerThe leaked Anthropic reasoning protocol. Running locally. Zero-drift coding with Strict Write Discipline and adaptive Cl