News for hackers. Stuff that matters.
Re:Language Models Are Super Mario: Absorbing Abilities from... by Drakim on Saturday April 06, 2024 @12:55PM (#39953669)
Do you know if there is anybody who has made it their mission to shrink models to extremes? It feels like the sorta thing somebody would get really obsessed with doing, akin to those people who make executable out of a few bytes or shrink network payloads.
Re:Language Models Are Super Mario: Absorbing Abilities from... by tomxor on Saturday April 06, 2024 @07:33PM (#39956674)
Ooh, parameter golfing! maybe I will get into ML one day after all.

On a similar note, I once found a paper that explained artificial neural networks computability from a very tiny ground up method, showing how few pieces you need to build a NAND gate... and of course once you have that you have everything.

Re:Language Models Are Super Mario: Absorbing Abilities from... by nico on Sunday April 07, 2024 @12:23AM (#39958174)
Some time ago, Andrej Karpathy posted a YouTube video about building GPT from scratch. Using the same dataset (a 1MB text file with all of Shakespeare’s works), the video goes from a very basic model to something like GPT2

What really stood out to me is that, while the typical wisdom is “more data -> better model”, in the video they use the exact same data for all the models, so all the gains are achieved only through improved algorithms and computation power

With that in mind, I wonder if at some point someone will figure out an algorithm for a model that can be trained on just an English dictionary and get a decent LLM-equivalent model that can have a basic conversation. Given the small size of the data (a dictionary), I assume the model would be pretty small and quite fast to run, even on older or smaller machines

Re:Language Models Are Super Mario: Absorbing Abilities from... by geon on Sunday April 07, 2024 @04:32AM (#39959281)
Do you mean dictionary as just a list of words in alphabetical order, or with detailed definitions for each word?
Re:Language Models Are Super Mario: Absorbing Abilities from... by nico on Sunday April 07, 2024 @10:37AM (#39961080)
Detailed definitions for each word, like a physical printed dictionary book
Re:Language Models Are Super Mario: Absorbing Abilities from... by geon on Sunday April 07, 2024 @04:34AM (#39959293)
Re:Language Models Are Super Mario: Absorbing Abilities from... by nico on Sunday April 07, 2024 @10:35AM (#39961056)
Yes, thank you, that’s the video
Re:Language Models Are Super Mario: Absorbing Abilities from... by abhgh on Sunday April 07, 2024 @01:46AM (#39958434)
Not for DL models, but I was exploring doing this in a specific setting (I've posted about this earlier). For small-sized models (for some reasonable definition of size, e.g., depth of a decision tree, or # trees in a gradient boosting forest, or # non-zero coefficients in a linear model), I realized that you could make them even smaller while retaining their accuracy by selectively presenting training data to them, i.e., ignore some training data points, repeat certain others. See [1] - the x-axis shows the original model size and the y-axis shows the model size obtained by this process at the same or better accuracy.

Interestingly, this meant that the conventional wisdom that the test and train distributions have to be identical for optimal held-out performance, is not true at small model sizes. It is true as models grow larger - and I was explicitly able to show this. See [2] - where the x-axis is model size, and the y-axis measures (on a scale of 0-1) how close is the optimal training distribution to the test distribution. The different lines are for different datasets.

These images are from the paper here [3]. I have a library too [4]; that is in need of updates - it works today as-is though, but please use the latest minor release if this of interest [5]!

[1] https://imgur.com/a/NheK49Z

[2] https://imgur.com/a/N53GeNI

[3] https://arxiv.org/pdf/1906.06852.pdf

[4] https://compactem.readthedocs.io

[5] As of writing the install command should be `pip install compactem==0.9.9rc2`