List your company

LLM Alignment on HH-RLHF 300 prompts

69.8Win/Tie Rate vs Vanilla (GPT-4o)CARDS
49.20854.55459.965.246Nov 5, 2025

Evaluation results

MethodLinks
CARDSModel=mistral-7b2025.1169.8
CARDSModel=llama-7b2025.1164.5
STARSModel=mistral-7b2025.1164.5
DPOModel=mistral-7b2025.1160.5
STARSModel=llama-7b2025.1160.2
Tree-bonModel=mistral-7b2025.1159.2
RAINModel=mistral-7b2025.1159
ARGSModel=mistral-7b2025.1158.8
DPOModel=llama-7b2025.1156.4
Tree-bonModel=llama-7b2025.1155.2
RAINModel=llama-7b2025.1155
ARGSModel=llama-7b2025.1154.8
Speculative-DecodingModel=mistral-7b2025.1150.4
Speculative-DecodingModel=llama-7b2025.1150.2
Vanilla LLMModel=llama-7b2025.1150
Vanilla LLMModel=mistral-7b2025.1150