AIRI unveils OCC-RAG compact language models for contextual Q&A

The Optimal Cognitive Core (OCC) team from AIRI has developed OCC-RAG, a family of compact language models (SLMs) designed for contextual question answering. Unlike large models focused on encyclopedic knowledge, OCC-RAG prioritizes reasoning and analysis of provided context. The initial release includes two checkpoints, OCC-RAG-0.6B and OCC-RAG-1.7B, which perform comparably to or better than much larger general-purpose models while demonstrating superior faithfulness to context among models up to 32B parameters. AI

IMPACT These compact models offer efficient contextual Q&A capabilities, potentially reducing computational costs for specific AI applications.

RANK_REASON Release of new language models with performance benchmarks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Mastodon — mastodon.social →

• AIRI
• OCC-RAG
• Optimal Cognitive Core

• model release
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →