Tiny Recursive Model
Summary
Tiny Recursive Model tracks the tiny recursive reasoning model that simplifies HRM while preserving strong puzzle-task generalization.
Role In The Wiki
Use this page as the object card for the model or method. The source page carries the evidence details, limitations, and agenda mapping.
Relation To Foundation TSFM Agenda
Use the source-level agenda mapping in tiny-recursive-model-2025 rather than duplicating verdict rows here. At the entity level, this page should stay as the object card; source pages carry slot-level evidence, limitations, and missing pieces.
Evidence
- Less is More: Recursive Reasoning with Tiny Networks
- Comparative critique: Fixed-Point Reasoners compares against TRM and frames part of TRM’s hierarchy/deep-supervision behavior as a signal-propagation and compute-allocation question.
- Inference extension: Probabilistic Tiny Recursive Model adds recurrent Gaussian perturbations and reuses TRM’s supervised Q head to select among parallel latent trajectories without retraining the checkpoint.