dOPSD: On-Policy Self-Distillation for Diffusion Language Models
Open paper
Benchmark
Model
Rank
Results
code-generation-on-mbpp
Dream-7B-Instruct (dOPSD)
#89
Accuracy: 58.49
code-generation-on-humaneval
Dream-7B-Instruct (dOPSD)
#97
Pass@1: 56.71
Rank counts only results with a code link.