INSAIT-Institute/GRPO-Instruct-7B

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.6BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 12, 2026License:cc-by-nc-sa-4.0Architecture:Transformer Open Weights Featherless Exclusive Cold

INSAIT-Institute/GRPO-Instruct-7B is a 7.6 billion parameter code verifier model developed by INSAIT-Institute, fine-tuned from Qwen/Qwen2.5-7B-Instruct. This model is trained using Reinforcement Learning with Verifiable Rewards (RLVR) on the Aletheia-Train dataset, specifically designed to judge and rank candidate solutions for competitive programming problems. Its key differentiator is its training without intermediate thinking traces, directly emitting a verdict, making it efficient for automated code evaluation and as a reward function in RLHF/RLAIF for code generation.

Loading preview...

Model Overview

GRPO-Instruct-7B is a 7.6 billion parameter code verifier developed by INSAIT-Institute, built upon the Qwen/Qwen2.5-7B-Instruct base model. It is part of the Aletheia project, which explores efficient training recipes for robust code verifiers. This specific model is trained using Reinforcement Learning with Verifiable Rewards (RLVR), incorporating on-policy learning and negative samples, but notably without intermediate thinking traces. This design allows it to directly output a verdict on code solutions, making it a streamlined verifier.

Key Capabilities

  • Code Verification: Judges and ranks candidate solutions for competitive programming problems, identifying correct code snippets among several options.
  • Efficiency: Designed to operate without generating intermediate thought processes, offering a potentially faster verification pipeline.
  • RLVR Training: Leverages on-policy learning and negative samples from the Aletheia-Train dataset for robust performance.

Use Cases

  • RLHF / RLAIF: Can serve as a plug-and-play reward function for optimizing code generation policies in reinforcement learning setups.
  • Automated Evaluation: Suitable for use as an LLM-as-a-judge in various code-related tasks, providing automated assessment of code quality and correctness.
  • Research: Valuable for researchers studying the impact of different training components (like thinking traces, on-policy learning, and negative samples) on code verifier performance and large language model performance.

Finetunes
1 models