con-cord/Mod3-no-ref

Mod3-no-ref: Medical LLM-as-a-Judge

Mod3-no-ref is a specialized large language model (LLM) developed by con-cord, built upon the Gemma-3-4B architecture. This 4.3 billion parameter model is fine-tuned to function as an "LLM-as-a-Judge" within the medical domain.

Key Capabilities

• Medical Response Evaluation: The primary function of Mod3-no-ref is to evaluate the quality and accuracy of generated medical answers.
• Clinical Criteria Assessment: It assesses responses based on predefined clinical evaluation criteria, ensuring relevance and correctness.
• Reference-Free Operation: A key differentiator of this specific version is its ability to perform evaluations without requiring expert reference answers, streamlining the assessment process.

Good For

• Automated quality assurance of AI-generated medical content.
• Evaluating the performance of other medical LLMs or response systems.
• Use cases where expert reference answers are unavailable or difficult to obtain for evaluation.