Transformers
Safetensors
mmada
diffusion-language-model
dllm
MMaDA
multimodal
WINO
WINO-plus
custom_code
Instructions to use QinFFF/WINO-plus-MMaDA-8B-MixCoT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use QinFFF/WINO-plus-MMaDA-8B-MixCoT with Transformers:
# Load model directly from transformers import MMadaModelLM model = MMadaModelLM.from_pretrained("QinFFF/WINO-plus-MMaDA-8B-MixCoT", trust_remote_code=True, dtype="auto") - Notebooks
- Google Colab
- Kaggle
WINO-plus-MMaDA-8B-MixCoT
This repository contains the full merged 8B model weights of WINO+, initialized from MMaDA-8B-MixCoT.
WINO+ is a trajectory-injection method for diffusion large language models. It transfers the verified denoising order discovered by WINO into model parameters, enabling faster and higher-quality generation.
The model corresponds to the WINO+ method described in Roll Out and Roll Back: Diffusion LLMs are Their Own Efficiency Teachers.
- Downloads last month
- 30
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for QinFFF/WINO-plus-MMaDA-8B-MixCoT
Base model
Gen-Verse/MMaDA-8B-MixCoT