LLaDA

The 'LLaDA' dataset/benchmark is used to evaluate the performance of diffusion large language models (DLLMs) in terms of their quality and efficiency during parallel generation tasks.

Papers using LLaDA (5)

dOPSD: On-Policy Self-Distillation for Diffusion Language Models2026BlockServe: Block-Grained Continuous Batching for High-Throughput Diffusion LLM Serving2026Roll Out and Roll Back: Diffusion LLMs are Their Own Efficiency Teachers2026Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding2025Sparse-dLLM: Accelerating Diffusion LLMs with Dynamic Cache Eviction2025