tttt111/test-tmp2
LFM2.5-1.2B-Instruct by Liquid AI is a 1.17 billion parameter instruction-tuned hybrid model with a 32,768 token context length, designed for efficient on-device deployment. It features extended pre-training on 28T tokens and multi-stage reinforcement learning, offering performance competitive with larger models. This model excels in agentic tasks, data extraction, and RAG, supporting fast inference on CPUs and mobile NPUs with low memory usage.
Loading preview...
LFM2.5-1.2B-Instruct: On-Device AI
LFM2.5-1.2B-Instruct is a 1.17 billion parameter instruction-tuned model developed by Liquid AI, part of the LFM2.5 family of hybrid models optimized for on-device deployment. It builds upon the LFM2 architecture with significantly extended pre-training (28 trillion tokens) and large-scale multi-stage reinforcement learning.
Key Capabilities & Features
- On-Device Performance: Designed for fast edge inference, achieving 239 tok/s decode on AMD CPU and 82 tok/s on mobile NPU, while operating under 1GB of memory.
- Broad Compatibility: Day-one support for
llama.cpp, MLX, and vLLM, with various quantized formats available (GGUF, ONNX, MLX). - Multilingual Support: Supports English, Arabic, Chinese, French, German, Japanese, Korean, and Spanish.
- Tool Use: Integrates function calling capabilities, allowing for structured interaction with external tools via a ChatML-like format.
- Competitive Benchmarks: Demonstrates strong performance against other sub-2B models across benchmarks like GPQA, MMLU-Pro, IFEval, and BFCLv3.
Ideal Use Cases
- Agentic Tasks: Well-suited for applications requiring autonomous decision-making or task execution.
- Data Extraction: Effective for extracting specific information from text.
- Retrieval Augmented Generation (RAG): Can be used to enhance generation with retrieved information.
- Edge Deployment: Optimized for deployment on devices such as mobile phones, laptops, IoT devices, and embedded systems due to its efficiency and low memory footprint.