Hardware Podcasts

160. AI Model Tests AI Models To Tell Us How They Work

The Next Byte Podcast · 6 Feb 2024 · 19m

What this episode covers

Hosts Daniel and Forbode discuss a team from MIT that has developed an 'Automated Interpretability Agent' (AIA) to analyze and explain the inner workings of other AI models. They detail how AIA uses pre-trained language models to probe neural networks, identifying strengths, weaknesses, and biases by testing individual neurons and composite behaviors. The hosts also cover the accompanying benchmark test, 'Find' (Function, Interpretation, and Description), which measures how well interpretation agents perform, noting that AIA is currently only about 50% effective on this standardized scale.

Most relevant to

makers and hobbyists

Topics

Mentioned

Companies
MIT, Wevolver
Products
AIA, Find, ChatGPT, Groq, Bard, Midjourney

Related episodes from other shows

Other shows on the same ground, closest first.

  • The Circuit · 48m

    Episode 109: Unpacking the Nuances of Deep Seek with Austin Lyons

    Austin Lyons, founder of ChipStrat, joins Jay Goldberg to analyze Deep Seek, a self-funded AI lab from China that emerged from the quantitative hedge fund High Flyer. They discuss Deep Seek's technical innovations, including its efficient Mixture of Experts (MoE) architecture and the R1 reasoning model, which demonstrated that high-performance AI is possible despite US chip export controls limiting access to NVIDIA H100s. The conversation explores how Deep Seek's training efficiency and open-weight models have shifted the industry focus from pure pre-training scaling to inference optimization and post-training techniques.

    AI & machine learningChips, FPGAs & semiconductors
    7 chapters

The summary, chapters and topics are generated from the episode’s audio by this directory, not written by The Next Byte Podcast. The episode itself belongs to its publisher.