LiquidAI has released LFM2.5-230M, a compact language model designed for on-device deployment. This model boasts 230 million parameters and is optimized for efficient inference on various hardware, including CPUs and edge devices, achieving speeds of up to 213 tokens/sec on a smartphone and 42 tokens/sec on a Raspberry Pi 5. It is particularly suited for agentic tasks due to its refinement through reinforcement learning and a large training budget of 19 trillion tokens, supporting a context length of 32,768 tokens. AI
IMPACT Enables more capable AI applications on resource-constrained devices, potentially accelerating the adoption of on-device AI for agentic tasks.
RANK_REASON Model release from a recognized AI lab (LiquidAI) with detailed technical specifications and performance metrics. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Hugging Face Trending Models →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
