You can deploy this model and start earning money today!

AbhilekhMeda/
qwen3-1.7b-math-prm

Views

No views yet

Qwen3-1.7B Math PRM

Training recipe:

Base model: Qwen/Qwen3-1.7B
Training data: raw step-level labels from Mai0313/prm800k (prm800k/data/phase2_train.jsonl)
Evaluation: Qwen/ProcessBench
Format: Qwen PRM-style <extra_0> marker after each reasoning step, score at marker positions

This repo contains the training script for a discriminative process reward model that scores individual reasoning steps.