AnonyRepo
/

bidir-prm-llada-8b-gsm8k

process-reward-model

discrete-diffusion

Model card Files Files and versions

bidir-prm-llada-8b

Bidirectional PRM on LLaDA-8B-Base (cross-backbone replication). Used in Appendix F; achieves ~0.32 accuracy on GSM8K under PRM-Guided K=8 (all configs). Base model: GSAI-ML/LLaDA-8B-Base.

Downloads last month: -

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for AnonyRepo/bidir-prm-llada-8b-gsm8k

Base model

GSAI-ML/LLaDA-8B-Base

Adapter

(12)

this model