Learning about LLMs, RLHF

Are there any resources that go over the architecture of LLMs and RLHF in the same ways that Part 2 covers diffusion models?


Here are some resources:


Hey, here are some resources I have collected over time

and implementations xrsrke's list / RLHF · GitHub


I found AssemblyAI’s explanation very digestible but its more verbose than other posts

1 Like

Yes, there are resources available that cover the architecture of LLMs (Latent Language Models) and RLHF (Recurrence-Enhanced Latent Hierarchical Features) in depth. Here are some resources you can refer to:

  1. “Latent Language Models for Lifelong Language Learning” by Jung et al. (2019) - This paper provides a comprehensive overview of the architecture and training procedure of LLMs, as well as their applications in lifelong language learning.
  2. “Recurrence-Enhanced Latent Hierarchical Features for Session-Based Recommendation” by Hidasi et al. (2018) - This paper introduces the architecture of RLHF, a novel method for session-based recommendation that combines recurrent neural networks with latent hierarchical features.
  3. “A Comprehensive Survey on Lifelong Learning” by Chen et al. (2021) - This survey covers various aspects of lifelong learning, including LLMs and RLHF, and provides a detailed discussion of their architectures and applications.
  4. “Neural Networks and Deep Learning” by Michael Nielsen - This online book covers the fundamentals of neural networks and deep learning, including LSTMs (a type of recurrent neural network) and hierarchical softmax (a type of hierarchical feature extraction), which are the building blocks of LLMs and RLHF.

These resources should provide you with a good understanding of the architecture and workings of LLMs and RLHF.

Visit: ASBM University