Podcast Episodes

Can Large Language Models Extract Customer Needs as well as Professional Analysts?

The paper investigates LLMs for extracting customer needs from reviews. Evaluations conducted with a professional marketing consulting firm. SFT LLMs…

1 year, 5 months ago

View Episode

Spurlens: finding spurious correlations in Multimodal llms

MLLMs exploit spurious correlations, affecting robustness and generalization The paper introduces SpurLens to identify and measure spurious cuesVario…

1 year, 5 months ago

View Episode

Architectural and Inferential Inductive Biases For Exchangeable Sequence Modeling

Autoregressive models effectively model exchangeable sequences and uncertainty from missing data The paper critiques single-step generation's limitat…

1 year, 5 months ago

View Episode

Improving test-time search with backtrack- Ing Improving test-time search with backtrack- Ing against in-context value verifiersagainst in-context value verifiers

Test-time verifiers improve reasoning performance by guiding solution chains Inefficient searches can arise from overlapping solutions and incorrect …

1 year, 5 months ago

View Episode

Adaptive elicitation of latent information Using natural language

The paper proposes an adaptive elicitation framework for reducing uncertainty It utilizes large language models for strategic information gatheringTh…

1 year, 5 months ago

View Episode

Document Valuation in LLM Summaries: A Cluster Shapley Approach

The paper addresses document valuation in LLM-generated summaries using Shapley valuesIt introduces the Cluster Shapley algorithm to enhance efficien…

1 year, 5 months ago

View Episode

s1: simple test time scaling

Test-time scaling improves language model performance using extra computeA dataset of 1,000 questions was curated for validationBudget forcing contro…

1 year, 5 months ago

View Episode