exdysa/Flux.1-lite-8B-MLX-Q4

Original Model Link : https://huggingface.co/Freepik/flux.1-lite-8B

name: flux.1-lite-8B-MLX-Q4
base_model: Freepik/flux.1-lite-8B
license: other
pipeline_tag: text-to-image
tasks :
 - text-to-image
 - image-generation
language: en
datasets: freepik
funded_by: Fal
hardware_type: 64 H100s
cloud_region :
 - Western Europe
 - East/Midwstern US
cloud_provider: Nebius
hours_used: 2880
get_started_code: uvx --from mflux mflux-generate --base-model dev --steps 50 --guidance 4.0 --seed 10 --width 1024 --height 1024 -q 4  --model exdysa/flux.1-lite-8B-MLX-Q4 --prompt 'Test Prompt' --prompt 'Test prompt'
library_name: mlx
tags:
- flux-1.schnell
- flux
- mlx
- apple
- Fal

flux.1-lite-8B

flux.1-lite-8B is an 8 billion parameter image synthesis model, a Diffusion Transformer derived from black-forest-labs/Flux.1-dev. The training data for this model originates from ~100k licensed stock images from Freepik.com. Separate self and cross attention layers, embedding extraction from intermediate T5 layers, and rotary positional embedding processes enabled an efficient model construction while maintaining quality performance. This model excels with illustrative and vector styles in particular.

MLX is a framework for METAL graphics supported by Apple computers with ARM M-series processors (M1/M2/M3/M4)

Generation using uv https://docs.astral.sh/uv/**:

uvx --from mflux mflux-generate --base-model dev --steps 50 --guidance 4.0 --seed 10 --width 1024 --height 1024 -q 4  --model exdysa/flux.1-lite-8B-MLX-Q4 --prompt 'Test Prompt'

Generation using pip:

pipx --from mflux mflux-generate --base-model dev --steps 50 --guidance 4.0 --seed 10 --width 1024 --height 1024 -q 4  --model exdysa/flux.1-lite-8B-MLX-Q4 --prompt 'Test Prompt'