# Efficient GPU inference
Omnigen2 Transformer DF11
The lossless DFloat11 compressed version of OmniGen2/OmniGen2, with the model size reduced by 32% while maintaining bit-level identical output and supporting efficient GPU inference.
Text-to-Image
O
DFloat11
593
2
FLUX.1 Canny Dev DF11
A version of black-forest-labs/FLUX.1-Canny-dev that uses the DFloat11 format for lossless compression, which can reduce GPU memory consumption by approximately 30% while maintaining bitwise identical outputs to the original model.
Text-to-Image
F
DFloat11
424
1
Labse Ru Sts
MIT
High-quality Russian sentence embedding BERT model, optimized based on cointegrated/LaBSE-en-ru, suitable for semantic text similarity tasks
Text Embedding
Transformers Other

L
sergeyzh
4,650
6
Featured Recommended AI Models