Evolutionary Task Discovery: Advancing Reasoning Frontiers via Skill Composition and Complexity Scaling

TL;DR

EvoTD introduces a structured evolutionary framework for data synthesis, enhancing reasoning capabilities of LLMs by systematically expanding task diversity and complexity.

Contribution

The paper presents EvoTD, a novel evolutionary approach with structured operators and a dynamic filter to improve reasoning by systematically generating diverse, complex tasks.

Findings

EvoTD achieves significant reasoning improvements across various models.

Structured evolutionary operators enhance task diversity and complexity.

Code is publicly available at https://github.com/liqinye/EvoTD.

Abstract

The reasoning frontier of Large Language Models (LLMs) has advanced significantly through modern post-training paradigms (e.g., Reinforcement Learning from Verifiable Rewards (RLVR)). However, the efficacy of these methods remains fundamentally constrained by the diversity and complexity of the training data. One practical solution is data synthesis; yet, prevalent methods relying on unstructured mutation or exploration suffer from homogeneity collapse, failing to systematically expand the reasoning frontier. To overcome this, we propose Evoutionary Task Discovery (EvoTD), a framework that treats data synthesis as a directed search over a dual-axis manifold of Algorithmic Skills and Complexity Attributes. We introduce structured evolutionary operators to navigate this space: a Crossover operator that synthesizes novel skill compositions to enhance diversity, and a Parametric Mutation operator that scales structural constraints (e.g., input size, tree depth) to drive robust generalization. Crucially, we integrate a dynamic Zone of Proximal Development filter, ensuring tasks lie within the learnable region of the model. Empirically, EvoTD delivers substantial reasoning gains that generalize consistently across model architectures, pretraining regimes, and scales, demonstrating that structured evolutionary curricula can effectively support reasoning improvement. We release our code on https://github.com/liqinye/EvoTD.