HYU-NLP-EVAL/qwen3-4b-rar-medicine-onlinerubrics-seed11-step-034

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:4BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 8, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

HYU-NLP-EVAL/qwen3-4b-rar-medicine-onlinerubrics-seed11-step-034 is a 4 billion parameter intermediate policy model derived from Qwen/Qwen3-4B-Instruct-2507, trained using dynamic OnlineRubrics-Every GRPO. This model represents a specific historical policy state from a Phase-1 audit, with thinking capabilities disabled. It is intended for research use only, specifically for evaluating policy states in reinforcement learning from human feedback (RLHF) contexts, and is not validated for clinical decision-making or any downstream medical applications.

Loading preview...

HYU-NLP-EVAL/qwen3-4b-rar-medicine-onlinerubrics-seed11-step-034 Overview

This model is an intermediate policy checkpoint from a dynamic OnlineRubrics-Every GRPO training process, distinct from static-rubric GRPO. It is based on the Qwen/Qwen3-4B-Instruct-2507 model, with its 'thinking' capabilities explicitly disabled. This specific checkpoint, identified as 'step 34, seed 11', represents a historical policy state used during a Phase-1 audit.

Key Characteristics

  • Policy Checkpoint: Represents an intermediate state from a reinforcement learning from human feedback (RLHF) training run.
  • Base Model: Built upon Qwen/Qwen3-4B-Instruct-2507, a 4 billion parameter instruction-tuned model.
  • Training Method: Utilizes dynamic OnlineRubrics-Every GRPO, a method for policy optimization.
  • Thinking Disabled: The model's internal 'thinking' mechanism is turned off in this version.
  • Research Use Only: Explicitly stated for research purposes; no medical capability or safety claims are made.

Good For

  • RLHF Research: Investigating the evolution of policies within dynamic OnlineRubrics-Every GRPO training.
  • Auditing Policy States: Analyzing specific historical checkpoints from RLHF training runs.
  • Understanding Policy Development: Studying how models behave at different stages of policy optimization.

Important Note: This model is not validated for clinical decision-making or any medical applications. It is provided as a research artifact to understand policy development, not for practical deployment in sensitive domains.