AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Human-like Few-Shot Learning via Bayesian Reasoning over Natural Language

Kevin Ellis

NeurIPS 2023 · Large language models › Language, safety and interpretability · approach: Faithfulness of LLM explanations · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Tune the prior to human judgmentsHuman-like Bayesian model · Logical conceptsR²Improved
Use natural-language hypotheses instead of Python hypotheses (ablation)Number Game model · Number Gamefit to human judgmentsImproved
Remove the data-conditioned proposal distribution and sample hypotheses from the prior (ablation)Number Game model · Number Gamefit to human judgmentsHurt
Replace importance sampling with deduplicating proposals (ablation)Human-like Bayesian model · Number Gamefit to human judgmentsNo clear effect
Replace Code-Davinci-002 with LLaMA2 70B as the likelihood model (ablation)Human-like Bayesian model · Number Gamefit to human judgmentsNo clear effect
Replace Code-Davinci-002 with LLaMA2 70B as the proposal distribution (ablation)Human-like Bayesian model · Number Gamefit to human judgmentsMixed
Use both concept-type-specific prompts instead of the first-order prompt for every concept (ablation)Human-like Bayesian model · Logical conceptsR²: .80 -> .81Improved
Use natural-language concepts with the learned prior rather than the BPL baselineLogical concept learning · Logical conceptsfit to human judgmentsImproved
Use Bayesian mechanics to reweight language-model proposalsGPT-4 concept proposals · Logical conceptsaccuracyImproved
Use the natural-language hypothesis space rather than the BPL baseline's fixed primitivesLogical concept learning · New majority-color and least-common-color conceptslearnabilityImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy