Eric Jang: Tips for Training Likelihood Models

This is a tutorial on common practices in training generative models that optimize likelihood directly, such as autoregressive models and ...

This is a tutorial on common practices in training generative models that optimize likelihood directly, such as autoregressive models and normalizing flows. Deep generative modeling is a fast-moving field, so I hope for this to be a newcomer-friendly introduction to the basic evaluation terminology used consistently across research papers, especially when it comes to modeling more complicated distributions like RGB images. This is a more in-depth version of the tutorial lecture I gave on normalizing flows at ICML. This tutorial discusses the most mathematically straightforward of generative…

saved by

• SUJASH AGRAWAL

related reading

• What are Diffusion Models? | Lil'Loglilianweng.github.io
• GenAI Handbookgenai-handbook.github.io
• The Principles of Diffusion Modelsarxiv.org
• The Principles of Diffusion Modelsarxiv.org
• [1809.09087] Implicit Maximum Likelihood Estimationarxiv.org
• What are Diffusion Models?lilianweng.github.io
• Denoising Density Estimatorsarxiv.org
• Yang Songyang-song.net
• Generative Modeling by Estimating Gradients of the Data Distribution | Yang Songyang-song.github.io
• Musings on typicality – Sander Dielemansander.ai
• Alexi Gladstone | Explorative Modeling -- Unlocking a Third Pretraining Axis and End-to-End Generationalexiglad.github.io
• Generative Modeling via Driftingarxiv.org