136 likesNew · 4d old
1.7B Model Packs 256K Context Into a 1 GB Phone
What is LFM3-1.7B?
LFM3-1.7B is a text-generation model that fits on a phone or laptop thanks to 4-bit quantisation and a 1 GB memory footprint. It provides a 256-thousand token window and is tuned for local agents and tool-calling.
Who it is for
Developers building edge AI assistants or domain-specific bots can run it locally without cloud costs and fine-tune it on a single GPU.
How to try it
Download the GGUF checkpoint and run with llama.cpp: `./main -m LFM3-1.7B.gguf -c 256000`