SoloQ: Calibration-Free Quantization for Diffusion Language Models

Open paper
Benchmark Model Rank Results
code-generation-on-humanevalSoloQ-N (Fast-dLLM v2-7B)–Pass@1: 59.1
mathematical-reasoning-on-minerva-mathSoloQ-C (Fast-dLLM v2-7B)–Accuracy: 54.86
question-answering-on-piqaSoloQ-N (LLaDA-1.5-8B)–Accuracy: 74.76

Rank counts only results with a code link.