Neural Networks Unpacked

A trained neural network naturally prioritizes complex information, allowing for a more controlled focus on desired features in embedding space. By analyzing the geometry of large language models, key characteristics can be identified, leading to insights on toxicity detection and generation. The approach simplifies the understanding of how prompts interact with model layers, revealing essential features that define their behavior.

In this clip
From this podcast
Machine Learning Street Talk (MLST)
Want to Understand Neural Networks? Think Elastic Origami! - Prof. Randall Balestriero
Related Questions
- What techniques are used with large language models (LLMs) in the episode Dodging Latent Space Detectors: Obfuscated Activation Attacks with Luke Bailey, Erik Jenner, an... and the clip Understanding Model Activations?
- How are large language models (LLMs) trained as discussed in the episode Mindscape 280 | François Chollet on Deep Learning and the Meaning of Intelligence and the clip Token Representations?

Neural Networks Unpacked

In this clip

From this podcast

Machine Learning Street Talk (MLST)

Want to Understand Neural Networks? Think Elastic Origami! - Prof. Randall Balestriero

Related Questions

What techniques are used with large language models (LLMs) in the episode Dodging Latent Space Detectors: Obfuscated Activation Attacks with Luke Bailey, Erik Jenner, an... and the clip Understanding Model Activations?

How are large language models (LLMs) trained as discussed in the episode Mindscape 280 | François Chollet on Deep Learning and the Meaning of Intelligence and the clip Token Representations?