WHAT-THE-JEV / EXPERIMENTS
Experiments
Benchmarks and investigations, with data, results, and analysis.
1. LLM Benchmarks: Academic Ability
01GPQA Diamond Graduate-Level Science
Explore JEV's performance on GPQA Diamond graduate-level science questions.
2. LLM Benchmarks: Social Commonsense and Bias
04BBQ Bias-Sensitive QA
Explore JEV's performance on BBQ's bias-sensitive questions, and whether it leans toward stereotypes.
4. Classical Machine Learning Prediction: Regression and Classification
09Boston Housing: Prediction and Comparison
Explore whether JEV can predict Boston suburb housing prices, and compare price levels across suburbs.
5. LLM Post-Training Annotation
11DPO Preference Annotation
Explore JEV's performance on preference annotation for DPO training data.