RecursiveMAS/Sequential-Light-Solver-Qwen2.5-Math-1.5B

TEXT GENERATIONConcurrency Cost:1Model Size:1.5BQuant:BF16Ctx Length:32kPublished:Apr 28, 2026License:mitArchitecture:Transformer Open Weights Cold

RecursiveMAS/Sequential-Light-Solver-Qwen2.5-Math-1.5B is a 1.5 billion parameter model developed by RecursiveMAS, built upon the Qwen2.5-Math-1.5B base. This model functions as a role-specific Solver Agent within the RecursiveMAS multi-agent framework, designed for sequential collaboration. It is optimized to produce final responses based on refined planning and critique information from other agents, rather than for standalone plain-text generation.

Loading preview...

Model Overview

This model, Sequential-Light-Solver-Qwen2.5-Math-1.5B, is a 1.5 billion parameter agent developed by RecursiveMAS. It is built on the Qwen2.5-Math-1.5B base and operates within the novel RecursiveMAS multi-agent framework, which scales agent collaboration through latent-space recursion. Unlike general-purpose language models, this is a role-specific agent.

Key Capabilities & Design

  • Solver Agent Role: Specifically designed to generate final responses within the RecursiveMAS framework.
  • Sequential-Light Collaboration: Functions in a sequential collaboration style, processing refined planning and critique information from other agents.
  • RecursiveMAS Framework: Integrates into a system where heterogeneous agents iteratively exchange, refine, and evolve their latent states across recursion rounds.
  • Not Standalone: This checkpoint is not intended for standalone plain-text generation but as a component of a larger multi-agent system.

When to Use This Model

  • Multi-Agent System Development: Ideal for researchers and developers implementing or experimenting with the RecursiveMAS framework.
  • Agent-Based Problem Solving: Suitable for use cases requiring a dedicated 'solver' component within a structured multi-agent collaboration pipeline.
  • Mathematical Reasoning Tasks: Given its Qwen2.5-Math base, it is likely to perform well in mathematical contexts when integrated into the RecursiveMAS system.

For detailed usage and integration instructions, refer to the RecursiveMAS GitHub repository and the associated research paper.