Advancing Block Diffusion Language Models for Test-Time Scaling TDAR-8B-Thinking is a state-of-the-art Block Diffusion Language Model (BDLM) designed for efficient test-time scaling on complex reasoning tasks. Built on Qwen3-8B architecture, it achieves 3.37× speedup over autoregressive baselines while maintaining superior reasoning quality. — 🚀 Bounded Adaptive Confidence Decoding (BACD): Dynamically adapts denoising process based on local difficulty signals — 💡 Think Coarse, Critic Fine (TCCF): Allocates computation based on functional roles in reasoning trajectories — 📈 Progressive Block Size Extension: Trained with gradually increasing block sizes (B=4→64) for optimal efficiency We use LMDeploy 0.10.2 with modifications for Bounded Adaptive Confidence Decoding support. > Note: This is a minimal setup for inference only. For full installation including training and evaluation dependencies, please refer to our comprehensive Installation Guide on GitHub. The following example shows how to quickly load the model and run inference end-to-end with BACD (Bounded Adaptive Confidence Decoding) for optimal speed-quality trade-off: We comprehensively evaluate TDAR on 6 diverse reasoning…
Модальности:
Генерация текста
Задача: Генерация текста
Автор: lulululuyi
Теги: sdar, custom_code, en
Лайков: 4 | Загрузок: 9
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.
