Open-ended text generation on WildBench
-1.7ScoreCTC-trained MDLM
Evaluation Results
| Method | Links | |
|---|---|---|
| CTC-trained MDLMTraining Objective=CTC + CE, Base Model=LLaDA-8B-Instruct2026.01 | -1.7 | |
| LLaDA-1.5Base Model=LLaDA-1.52026.01 | -2.5 | |
| LLaDA-8B-InstructBase Model=LLaDA-8B-Instruct2026.01 | -3.3 | |
| CE-only baselineTraining Objective=CE Only, Base Model=LLaDA-8B-Instruct2026.01 | -4.5 |