alexweberk

veekaybee

Anti-hype LLM reading list

Goals: Add links that are reasonable and good explanations of how stuff works. No hype and no vendor content if possible. Practical first-hand accounts of models in prod eagerly sought.

Foundational Concepts

Pre-Transformer Models

yoavg

Reinforcement Learning for Language Models

Yoav Goldberg, April 2023.

Why RL?

With the release of the ChatGPT model and followup large language models (LLMs), there was a lot of discussion of the importance of "RLHF training", that is, "reinforcement learning from human feedback". I was puzzled for a while as to why RL (Reinforcement Learning) is better than learning from demonstrations (a.k.a supervised learning) for training language models. Shouldn't learning from demonstrations (or, in language model terminology "instruction fine tuning", learning to immitate human written answers) be sufficient? I came up with a theoretical argument that was somewhat convincing. But I came to realize there is an additional argumment which not only supports the case of RL training, but also requires it, in particular for models like ChatGPT. This additional argument is spelled out in (the first half of) a talk by John Schulman from OpenAI. This post pretty much

seanjtaylor

rmcelreath

# functions for plotting garden of forking data plots

library(rethinking)

polar2screen <- function( dist, origin, theta ) {
  ## takes dist, angle and origin and returns x and y of destination point
  vx <- cos(theta) * dist;
  vy <- sin(theta) * dist;
  c( origin[1]+vx , origin[2]+vy );
}

nylki

do androids dream of cooking?

The following recipes are sampled from a trained neural net. You can find the repo to train your own neural net here: https://github.com/karpathy/char-rnn Thanks to Andrej Karpathy for the great code! It's really easy to setup.

The recipes I used for training the char-rnn are from a recipe collection called ffts.com And here is the actual zipped data (uncompressed ~35 MB) I used for training. The ZIP is also archived @ archive.org in case the original links becomes invalid in the future.

jboner

Latency Comparison Numbers (~2012)
----------------------------------
L1 cache reference                           0.5 ns
Branch mispredict                            5   ns
L2 cache reference                           7   ns                      14x L1 cache
Mutex lock/unlock                           25   ns
Main memory reference                      100   ns                      20x L2 cache, 200x L1 cache
Compress 1K bytes with Zippy             3,000   ns        3 us
Send 1K bytes over 1 Gbps network       10,000   ns       10 us
Read 4K randomly from SSD*             150,000   ns      150 us          ~1GB/sec SSD