Self-consistency: trusting the majority
Asking the same question K times and taking the majority answer can make a model that is usually wrong almost flawless. On one condition.
4 steps
225 XP
A free account is needed
Start the lesson →
Sources
- Wang, X. et al. 2023 · Self-Consistency Improves Chain of Thought Reasoning in Language Models · ICLR 2023
- Cobbe, K. et al. 2021 · Training Verifiers to Solve Math Word Problems · arXiv:2110.14168
- Snell, C. et al. 2024 · Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters · arXiv:2408.03314
ML Academy · an interactive machine learning course that runs in your browser ·
All lessons