ai-forever/pollux-judge-7b
TEXT GENERATIONConcurrency Cost:1Model Size:7.6BQuant:FP8Ctx Length:32kPublished:Jun 5, 2025License:mitArchitecture:Transformer0.0K Open Weights Cold

ai-forever/pollux-judge-7b is a 7.6 billion parameter decoder-based generative language model developed by ai-forever, specifically designed for evaluating the quality of other LLMs' responses in Russian. Built on the t-tech/T-lite-it-1.0 architecture, it assesses answer quality given instructions, criteria, and rubrics, providing automated performance evaluation for Russian-language tasks. The model predicts both numerical scores and detailed textual rationales, excelling in single-criterion evaluation for Russian LLM assessment.

Loading preview...