munin-lfm25-1b2-4bit-mlx

On-device summarization tier for Munin.

This is an unmodified mirror of LiquidAI/LFM2.5-1.2B-Instruct-MLX-4bit at revision 7ccafdb04c36936f4f1c4685198c6c9a40275932 — Liquid AI's own MLX 4-bit build of LFM2.5-1.2B-Instruct. No re-quantization or conversion was performed here. The only added file is manifest.json, which Munin uses to verify the bundle's SHA-256 hashes after download.

Mirrored so the app has a manifest to verify against and a stable revision to pin. Prefer the upstream repo for any other use.

License

Governed by the LFM Open License v1.0, redistributed here as LICENSE. It is Apache-2.0-based but is not an OSI-approved open-source licence: commercial use is free only while the using entity is under $10M USD annual revenue. At or above that threshold a separate licence from Liquid AI is required. Review it before depending on this model commercially.

Downloads last month: -

Safetensors

Model size

0.2B params

Tensor type

BF16

U32

MLX

Hardware compatibility

4-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for asksolz/munin-lfm25-1b2-4bit-mlx

Base model

LiquidAI/LFM2.5-1.2B-Base

Finetuned

LiquidAI/LFM2.5-1.2B-Instruct

Quantized

LiquidAI/LFM2.5-1.2B-Instruct-MLX-4bit

Quantized

(1)

this model