OPDLM-MATH-4B-Thinking
OPDLM-MATH-4B-Thinking is an On-Policy Distillation Language Model (OPDLM) — a block-diffusion language model (block size 4, 4 denoising steps per block) post-trained for mathematical reasoning. It is built on a block-diffusion adaptation of Qwen/Qwen3-4B (architecture a2d-qwen3).
This is the thinking variant: long-context (8k) math post-training with an explicit chain-of-thought thinking block enabled at inference.
arXiv report: arxiv.org/abs/2606.06712
Usage
This model uses custom modeling code; load with trust_remote_code=True. Generation is block-diffusion (non–left-to-right), so use the project's inference utilities (block_size=4, denoising_steps_per_block=4) rather than vanilla model.generate.
from transformers import AutoModel, AutoTokenizer
tok = AutoTokenizer.from_pretrained("divelab/OPDLM-MATH-4B-Thinking", trust_remote_code=True)
model = AutoModel.from_pretrained("divelab/OPDLM-MATH-4B-Thinking", trust_remote_code=True)