List your company

Reward Model Transfer on Anthropic Harmless (AHar)

4.62AOGANCHOR
1.21922.10212.9853.8679May 25, 2026

Evaluation results

MethodLinks
ANCHORTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=PKU, Evaluation Setting=Zero-shot transfer2026.054.62——4.62
ANCHORTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=RAIL, Evaluation Setting=Zero-shot transfer2026.052.62——2.62
NaiveTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=RAIL, Evaluation Setting=Zero-shot transfer2026.052.44——2.44
NaiveTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=PKU, Evaluation Setting=Zero-shot transfer2026.052.26——2.26
NaiveTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=RAIL, Evaluation Setting=Zero-shot transfer2026.051.53——1.02
ANCHORTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=PKU, Evaluation Setting=Zero-shot transfer2026.051.43——1.43
ANCHORTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=RAIL, Evaluation Setting=Zero-shot transfer2026.051.4——1.4
NaiveTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=PKU, Evaluation Setting=Zero-shot transfer2026.051.35——1.35
ANCHORTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=AHar, Evaluation Setting=In-distribution2026.05—134.53.39—
ANCHORTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=AHar, Evaluation Setting=In-distribution2026.05—160.91.48—
NaiveTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=AHar, Evaluation Setting=In-distribution2026.05—122.63.09—
NaiveTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=AHar, Evaluation Setting=In-distribution2026.05—160.91.48—