rubricreward/R3-Qwen3-8B-14k
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:32kPublished:May 14, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

rubricreward/R3-Qwen3-8B-14k is an 8 billion parameter reward model, part of the R3 (Robust Rubric-Agnostic Reward Models) family, fine-tuned from Qwen3-8B. It is trained on a diverse dataset covering classification, preference optimization, and question answering tasks, with examples including instructions, inputs, responses, evaluation rubrics, scores, and reasoning. This model specializes in evaluating responses based on detailed rubrics, providing scores and reasoning for assessment tasks.

Loading preview...