List your company

Reward Modeling on PPE Pref

67.7AccuracyConsistRM (Qwen3-8B)
53.76457.3826164.618Apr 8, 2026

Evaluation results

MethodLinks
ConsistRM (Qwen3-8B)Base Model=Qwen3-8B, Training Method=CAAR + CACR, GT=False2026.0467.7
DeepSeek-GRM-27BBase Model=DeepSeek-27B, GT=True2026.0467.2
Qwen3-8B + RFTBase Model=Qwen3-8B, Training Method=RFT, GT=True2026.0465.4
ConsistRM (Qwen3-4B)Base Model=Qwen3-4B, Training Method=CAAR + CACR, GT=False2026.0465.1
Qwen3-8B + SFTBase Model=Qwen3-8B, Training Method=SFT, GT=True2026.0465
Qwen3-8B + TTRLBase Model=Qwen3-8B, Training Method=TTRL, GT=False2026.0465
Qwen3-8B + CAARBase Model=Qwen3-8B, Training Method=CAAR, GT=False2026.0464.8
Qwen3-4B + RFTBase Model=Qwen3-4B, Training Method=RFT, GT=True2026.0464.7
Qwen3-4B + CAARBase Model=Qwen3-4B, Training Method=CAAR, GT=False2026.0464.1
Qwen3-8BBase Model=Qwen3-8B2026.0463.8
Qwen3-4BBase Model=Qwen3-4B2026.0463.3
Qwen3-4B + SFTBase Model=Qwen3-4B, Training Method=SFT, GT=True2026.0463.1
Qwen3-4B + TTRLBase Model=Qwen3-4B, Training Method=TTRL, GT=False2026.0461.8
J1-Llama-8BBase Model=Llama-8B, GT=True2026.0459.8
EvalPlanner-Llama-8BBase Model=Llama-8B, GT=True2026.0454.3