InternLM has introduced Intern-S2-397B, a new multimodal foundation model designed for scientific intelligence and long-horizon agents. This model enhances capabilities through a novel visual pre-training paradigm that directly processes scientific literature, improving data efficiency and reasoning. Intern-S2-397B also excels in scientific modality reasoning and generation by training on diverse scientific reinforcement learning tasks, and it demonstrates improved generalization in long-horizon tasks through connection with large-scale agentic reinforcement learning environments. AI
IMPACT This model's focus on scientific literature and long-horizon agents could advance AI capabilities in research and complex task execution.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
