LMSYS-Chat-1M

anchor

Findings

• IC-238LLMs fail to follow user preferences in zero-shot settings, with accuracy below 10% at 10 turns and near zero at 300 turns [source]
• IC-252GPT models produce harmful gender stereotypes at higher rates when user names imply a demographic group, with GPT-3.5 Turbo showing the highest rates and open-ended generation tasks most affected [source]
• IC-253Post-training reinforcement learning significantly reduces harmful gender stereotypes in GPT models, with the best-fit slope of 0.21 indicating post-RL models have far lower bias than pre-RL versions [source]
• IC-254GPT-4o Mini responses to female-sounding names systematically use simpler, more light-hearted, and less technical language compared to male-sounding names across multiple task domains [source]
• IC-255GPT-4o, Llama 3.1, and Claude models show varying correlation with human ratings when used as stereotype evaluators, with GPT-4o achieving the strongest gender correlation (ρ=0.86) but weaker racial correlations [source]
• IC-750Open-source models without safety training are significantly more vulnerable to jailbreak attacks than safety-aligned proprietary models [source]
• IC-752GPT-4's win rate over GPT-3.5-turbo is 52% on the top-50 most challenging prompts but only 22% on the bottom-50 easiest prompts [source]