Modelpublic

LiquidAI/LFM2.5-2.6B-MLX-5bit

sourceHugging Faceotherupdated 1mo agoView on Hugging Face

2likes270downloads

Model Card

<div align="center"> <img src="https://cdn-uploads.huggingface.co/production/uploads/61b8e2ba285851687028d395/2b08LKpev0DNEk6DlnWkY.png" alt="Liquid AI" style="width: 100%; max-width: 100%; height: auto; display: inline-block; margin-bottom: 0.5em; margin-top: 0.5em;" /> <div style="display: flex; justify-content: center; gap: 0.5em; margin-bottom: 1em;"> <a href="https://playground.liquid.ai/"><strong>Try LFM</strong></a> • <a href="https://docs.liquid.ai/lfm/getting-started/welcome"><strong>Docs</strong></a> • <a href="https://leap.liquid.ai/"><strong>LEAP</strong></a> • <a href="https://discord.com/invite/liquid-ai"><strong>Discord</strong></a> </div> </div> <br>

LFM2.5-2.6B-MLX-5bit

MLX export of LFM2.5-2.6B for Apple Silicon inference.

LFM2.5-2.6B is a compact multilingual model built on LiquidAI's hybrid architecture, combining convolutional and attention layers for efficient long-context processing.

Model Details

Property	Value
Parameters	2.6B
Precision	5-bit
Group Size	64
Size	1.76 GB
Context Length	131072

Use with mlx

bash

pip install mlx-lm

python

from mlx_lm import load, generate
from mlx_lm.sample_utils import make_sampler

model, tokenizer = load("LiquidAI/LFM2.5-2.6B-MLX-5bit")

response = generate(
    model,
    tokenizer,
    prompt="The capital of France is",
    max_tokens=100,
    sampler=make_sampler(temp=0.7),
    verbose=True,
)

Other Precisions

—LFM2.5-2.6B-MLX-bf16 (5.02 GB)
—LFM2.5-2.6B-MLX-8bit (2.67 GB)
—LFM2.5-2.6B-MLX-6bit (2.04 GB)
—LFM2.5-2.6B-MLX-5bit (1.76 GB)
—LFM2.5-2.6B-MLX-4bit (1.47 GB)
—LFM2.5-2.6B-MLX-mxfp8 (2.59 GB)
—LFM2.5-2.6B-MLX-mxfp4 (1.46 GB)
—LFM2.5-2.6B-MLX-nvfp4 (1.53 GB)

License

This model is released under the LFM 1.0 License.