List your company

Reward Model Transfer on UltraFeedback (UF)

7.93AOGANCHOR
-0.79561.46973.7356.0003May 25, 2026

Evaluation results

MethodLinks
ANCHORTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=H3, Evaluation Setting=Zero-shot transfer2026.057.93——7.93
Conf.-basedTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=H3, Evaluation Setting=Zero-shot transfer2026.057.64——5.52
SEAMTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=H3, Evaluation Setting=Zero-shot transfer2026.056.25——-9.42
SEAMTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=H3, Evaluation Setting=Zero-shot transfer2026.055.79——-5.8
ANCHORTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=AH, Evaluation Setting=Zero-shot transfer2026.055.73——5.73
ANCHORTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=AH, Evaluation Setting=Zero-shot transfer2026.055.61——5.61
SEAMTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=AH, Evaluation Setting=Zero-shot transfer2026.055.49——2.07
NaiveTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=AH, Evaluation Setting=Zero-shot transfer2026.055.15——5.15
NaiveTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=AH, Evaluation Setting=Zero-shot transfer2026.054.98——4.98
Conf.-basedTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=AH, Evaluation Setting=Zero-shot transfer2026.054.92——4.92
NaiveTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=H3, Evaluation Setting=Zero-shot transfer2026.054.8——4.8
Conf.-basedTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=AH, Evaluation Setting=Zero-shot transfer2026.054.74——4.74
ANCHORTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=H3, Evaluation Setting=Zero-shot transfer2026.054.69——4.69
Conf.-basedTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=H3, Evaluation Setting=Zero-shot transfer2026.054.4——2.1
NaiveTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=H3, Evaluation Setting=Zero-shot transfer2026.050.17——-2.89
SEAMTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=AH, Evaluation Setting=Zero-shot transfer2026.05-0.46——-13.5
ANCHORTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=UF, Evaluation Setting=In-distribution2026.05—97.92.32—
ANCHORTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=UF, Evaluation Setting=In-distribution2026.05—189.10.87—
Conf.-basedTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=UF, Evaluation Setting=In-distribution2026.05—41.80.99—
Conf.-basedTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=UF, Evaluation Setting=In-distribution2026.05—-39.1-0.18—
NaiveTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=UF, Evaluation Setting=In-distribution2026.05—73.41.74—
NaiveTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=UF, Evaluation Setting=In-distribution2026.05—-89.1-0.41—
SEAMTransfer Pair=Llama-3.2-1B-Instruct -> Llama-3.1-8B-Instruct, Source Domain=UF, Evaluation Setting=In-distribution2026.05—-280.6-6.65—
SEAMTransfer Pair=Qwen3-1.7B -> Qwen3-8B, Source Domain=UF, Evaluation Setting=In-distribution2026.05—-517.4-2.38—