INSAIT-Institute/GRPO-Think-14B-4k

TEXT GENERATIONPricing:Input $0.431 / Cached $0.0862 / Output $1.12Concurrent Unit Cost:1Model Size:14.8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 12, 2026License:cc-by-nc-sa-4.0Architecture:Transformer Open Weights Featherless Exclusive Cold

INSAIT-Institute/GRPO-Think-14B-4k is a 14.8 billion parameter GRPO-Think code verifier model developed by INSAIT-Institute, fine-tuned from DeepSeek-R1-Distill-Qwen-14B. This model is trained with Reinforcement Learning with Verifiable Rewards (RLVR) using a 4k reasoning-token budget, excelling at judging and ranking candidate solutions for competitive programming problems. It is designed to act as a plug-and-play reward function for code generation policy optimization and automated evaluation.

Loading preview...

Model Overview

GRPO-Think-14B-4k is a 14.8 billion parameter code verifier developed by INSAIT-Institute, fine-tuned from deepseek-ai/DeepSeek-R1-Distill-Qwen-14B. It is part of the Aletheia project, which investigates the effectiveness of Reinforcement Learning with Verifiable Rewards (RLVR) for code verifiers. This specific model is trained using the GRPO (Generative Reinforcement Pre-training with Optimization) approach, incorporating on-policy learning, negative samples, and intermediate thinking traces, with a 4k reasoning-token budget during training.

Key Capabilities

  • Code Verification: Judges and ranks candidate solutions for competitive programming problems, given a problem statement and multiple code snippets.
  • RLVR Training: Utilizes a comprehensive RLVR pipeline, including intermediate thinking traces, learning from negative samples, and on-policy training.
  • Scalable Performance: Research indicates that for larger models like this 14B variant, the 'thinking budget' (reasoning-token budget) becomes a vital factor in performance, and eliminating on-policy training can still yield comparable results.

Intended Uses

  • RLHF / RLAIF: Serves as a plug-and-play reward function for optimizing code generation policies.
  • Automated Evaluation: Can be used as an LLM-as-a-judge for various code-related tasks.
  • Research: Provides a controlled testbed for studying the impact of thinking traces, on-policy learning, and negative samples in training robust code verifiers.

Finetunes
1 models