CoolFace
Modelpublic

KoarAI/LFM2.5-350M-Thinking-0005-GGUF

sourceHugging Faceapache-2.0updated 26d agoView on Hugging Face
0likes842downloads
Model Card

<p align="center"> <img src="banner.png" alt="LFM2.5-350M-Thinking-0005 Banner" width="100%"> </p>

🧠 LFM2.5-350M-Thinking-0005-GGUF

Quantized GGUF binaries of KoarAI/LFM2.5-350M-Thinking-0005 for ultra-fast local inference on CPU, Vulkan, and Apple Metal via llama.cpp.

📦 Available Files

FilePrecisionSizeDescription
LFM2.5-350M-Thinking-0005-f16.ggufFP16~709 MBFull precision golden baseline
LFM2.5-350M-Thinking-0005-q8_0.ggufQ8_0~380 MBHigh-quality quantized balance

🚀 Quick Start (llama.cpp)

bash
llama-cli -m LFM2.5-350M-Thinking-0005-f16.gguf -p "<|im_start|>user\nСколько букв 'р' в слове 'трапеция'? Разбери пошагово.\n<|im_end|>\n<|im_start|>assistant\n" -n 512 --temp 0.6