Spaces:
Running
Apply for a GPU community grant: Company project
Collov-Labs/Monetico is a public text-to-image demo for the open Monetico model hosted on Hugging Face.
The model contains approximately 6 GB of weights and requires iterative diffusion inference. On CPU Basic, generation takes many minutes and the interactive demo is effectively unusable.
We request a Community GPU Grant so users can generate images within a reasonable interactive time. We have added Fast, Standard, and Quality presets; Fast is the default at 384x384 with 12 inference steps and uses about 86% less transformer work than the previous 512x512, 48-step default. The Space also uses inference-only execution, concurrency 1, dynamic ZeroGPU duration estimates, and automatic fp16 fallback on GPUs without native bfloat16 support.
An Nvidia L4 or equivalent allocation with native bfloat16 support and at least 24 GB VRAM would provide the best balance of compatibility and cost.
Model: https://huggingface.co/Collov-Labs/Monetico
Space: https://huggingface.co/spaces/Collov-Labs/Monetico