The Little Book of Reinforcement Learning

johnea | 11 days ago

Is this riffing on Strunk and Whites: The Elements of Style?

Often referred to as "The Little Book".

leoc | 11 days ago

Most likely not: “The Little Book of …” has been a publisher’s standby since the nineteenth century (at least).

AlexB138 | 11 days ago

There are several "Libellus de Miraculis" (Little Book of Miracles) of different saints from the 12th century!

reader9274 | 11 days ago

Love that masterpiece

relyks | 11 days ago

I'm assuming it's more in line with The Little Schemer series of books (https://felleisen.org/matthias/BTLS-index.html) or maybe the little book of deep learning (https://fleuret.org/francois/lbdl.html)?

barrenko | 11 days ago

Proobably the later.

fxwin | 11 days ago

definitely the latter, it is even referenced in the foreword:

> Its goal is not to be exhaustive, but rather minimalist and easy to read. For this reason, it follows the format of The Little Book of Deep Learning [Fleuret 2023]. Its tone, however, is closer to that of a blog post, as the book is built around a single narrative thread. Its structure broadly follows that of Sutton and Barto’s Reinforcement Learning: An Introduction [Sutton et al. 2018], which remains the canonical reference on the subject.

Exoristos | 11 days ago

No, it's _obviously_ homage to the Little Liddel[0].

0. https://archive.org/details/lexiconabridgedf00liddrich

sublinear | 11 days ago

My mind immediately went to the "Li'l Bastard General Mischief Kit" from The Simpsons.

https://imgur.com/zMTEE

rcyeh | 11 days ago

I thought the title was an homage to François Fleuret's Little Book of Deep Learning.

https://fleuret.org/francois/lbdl.html

tejtm | 11 days ago

The Little Schemer, The Little Typer, The Little Reasoner, The Little Proover The Little MLer ...

It has been going on for a while in Lispy land

verdverm | 11 days ago

This looks like a good pre-read for Nathan Lambert's https://rlhfbook.com/

newsomix9xl | 11 days ago

Real biological operant behavior isn't exactly trial and error learning.

Many factors shape and guide initial responses.

What I've noticed in some descriptions of models is the use of optimization for reinforcement to shape responses. In real organisms behavior may be controlled by short or long term outcomes, and may oscillate between this "optimization" based on schedules. This produces variability in the trials which can adjust behavior. Are we seeing these reinforcement models do this?

herodoturtle | 11 days ago

I found this comment/question deeply intriguing.

I’m no expert at this and was wondering what you meant by the following:

> In real organisms behavior may be controlled by short or long term outcomes, and may oscillate between this "optimization" based on schedules

Could you perhaps provide an example that would help me understand what you mean?

Thanks for the insightful comment either way.

newsomix9xl | 10 days ago

In humans this is often called impulsivity, the preference for smaller sooner outcomes. This is often seen in children and animals and in some adult human behavior.

An impulsive choice is not optimal. You can buy a cheaper pack of gum at Costco in a week or get one for three times the cost right now.

ainch | 11 days ago

There is a field of hierarchical RL in which the optimisation occurs over a range of time scales/abstraction. But I'm not aware of much practical success for these approaches so far.

programjames | 11 days ago

I skimmed through the book, and it's lacking the information theory foundations. For example, "trust region methods" come from maximizing the policy's relative entropy (to a reference policy) under a tournament system where high-scoring agents are exponentially likely to survive. In general, a reward is the negative bits it costs an environment to propagate an agent (multiplied by some temperature).

ainch | 11 days ago

Do you have a good source on this information theory framing? I don't remember it being covered in Sutton & Barto.

porridgeraisin | 11 days ago

It's just another way to frame it. It's as foundational as the many other ways to frame it. I'm not aware of any major insight you get specifically from this framing. Is there one?

janalsncm | 11 days ago

I wonder what Sutton thinks about some of the more recent innovations in RL like GRPO. In some ways it’s new, in other ways it’s an echo of RLOO.

porridgeraisin | 11 days ago

GRPO is policy gradient/PPO with your value function baseline monte carlo estimated using k rollouts. The only new thing is finding out it works well with binary rewards and LLM policies.

janalsncm | 10 days ago

It is a huge improvement to PPO because you don’t need a separate critic model which cuts memory costs in half and stabilizes training.

porridgeraisin | 10 days ago

Yes, but monte carlo estimating the critic model is not new.

Envwnger | 11 days ago

Should have named it little RL book.

laurensr | 11 days ago

This reminds me of the Little Book of Calm, discussed extensively in the Black Books TV series.

wpm | 11 days ago

Thank you for reminding me to rewatch Black Books.