Training Optimal Large Diffusion Language Models Jinjie Ni†, Qian Liu, Chao Du, Longxu Dou, Hang Yan, Zili Wang, Tianyu Pang, Michael Qizhe Shieh
](https://huggingface.co/Dream-org/Dream-v0-Base-7B)
We introduce LLaDA 1.5, a competitive large diffusion language model, trained by variance-reduced preference optimization (VRPO).
Current Diffusion Language Models (DLMs) have been studied at a smaller scale compared to their autoregressive (AR) counterparts and lack fair comparison on language modeling benchmarks. Additionally, training diffusion models from scratch at scale remains challenging. We propose adapting…
Les fonctionnalités principales de hkunlp/diffullama sont : Language Diffusion Models, Training and Alignment.
Les alternatives open-source à hkunlp/diffullama incluent : jinjieni/quokka — Training Optimal Large Diffusion Language Models Jinjie Ni†, Qian Liu, Chao Du, Longxu Dou, Hang Yan, Zili Wang,… ml-gsai/llada-1.5 — We introduce LLaDA 1.5, a competitive large diffusion language model, trained by variance-reduced preference… hkunlp/dream — ](https://huggingface.co/Dream-org/Dream-v0-Base-7B). jinjieni/megadlms — MegaDLMs. autonomousvision/mdpo — [[Paper]](https://arxiv.org/pdf/2508.13148) [[Project]](https://cli212.github.io/MDPO/). amap-ml/ar-map — Are Autoregressive Large Language Models Implicit Teachers for Diffusion Large Language Models? A comprehensive…