yoavg

Learning LLMs in 2025

So you know how the transformer works, and you know basic ML/DL, and you want to learn more about LLMs. One way to go is looking into the various "algorithmic" stuff (optimization algorithms, RL, DPO, etc). Lot's of materials on that. But the interesting stuff is (in my opinion at least) not there.

This is an attempt to collect a list of academic (or academic-like) materials that explore LLMs from other directions, and focus on the non-ML-algorithmic aspects.

Courses

• David Chiang's Theory of Neural Networks course.
• This is not primarily LLMs, but does have substantial section on Transformers. Formal/Theory. More of a book than a course.

roycoding

It's 2020 and I made an RSS feed for my blog (with Python)

Roy Keyes

21 May 2020 - This is a post on my blog. Grab the RSS feed here 😉

I was recently telling someone about all of the awesome blog posts I have sitting in my backlog, just waiting to actually be written. They asked if there was a way to subscribe or get notified when my next post was live. The answer was "watch my Twitter", which was admittedly unsatisfactory.

Coincidentally I have been seeing a number of recent articles on the desire for more independent blog content and a return to RSS-based syndication. I think this is all part of a broader push for a (return to a?) m

AndyShiue

I think I’ve figured out most parts of the cubical type theory papers; I’m going to take a shot to explain it informally in the format of Q&As. I prefer using syntax or terminologies that fit better rather than the more standard ones.

Q: What is cubical type theory?

A: It’s a type theory giving homotopy type theory its computational meaning.

Q: What is homotopy type theory then?

A: It’s traditional type theory (which refers to Martin-Löf type theory in this Q&A) augmented with higher inductive types and the univalence axiom.

kumarbhrgv

Improvising diversity of personalized recommendation systems

Recent Research papers:

• Improving Aggregate Recommendation Diversity Using Ranking-Based Techniques:

  107 Citations : IEEE Transactions on Knowledge and Data Engineering
  we introduce and explore a number of item ranking techniques that can generate substantially more diverse recommendations across all users while maintaining comparable levels of recommendation accuracy. Comprehensive empirical evaluation consistently shows the diversity gains of the proposed techniques using several real-world rating data sets and different rating prediction algorithms

• Recommendation Diversification Using Explanations: (Data Engineering, 2009. ICDE '09. IEEE 25th International Conference)

Traditionally, the problem is addressed through attribute-based diversification grouping items in the result set that share many common attributes (e.g., genre for movies) and selecting only a limited number of items from each group. It is, however,

gbaman

# An example to get the remaining rate limit using the Github GraphQL API.

import requests

headers = {"Authorization": "Bearer YOUR API KEY"}


def run_query(query): # A simple function to use requests.post to make the API call. Note the json= section.
    request = requests.post('https://api.github.com/graphql', json={'query': query}, headers=headers)
    if request.status_code == 200:

lorey

from bs4 import BeautifulSoup
from markdown import markdown
import re

def markdown_to_text(markdown_string):
    """ Converts a markdown string to plaintext """

    # md -> html -> text since BeautifulSoup can extract text cleanly
    html = markdown(markdown_string)

jbochi

loong

# Taken from https://forums.aws.amazon.com/thread.jspa?messageID=332091

sudo su -

cd /usr/local/bin
mkdir ffmpeg

cd ffmpeg
wget http://ffmpeg.gusari.org/static/64bit/ffmpeg.static.64bit.latest.tar.gz
tar -xzf ffmpeg.static.64bit.latest.tar.gz

aparrish

montenegrodr

from vowpalwabbit import pyvw

def to_vw(clf, text, str_labels):
    vw_example = str('{} |f {} '.format(str_labels, text))
    return clf.example(vw_example)

clf = vw = pyvw.vw(
    loss_function='logistic', oaa=2,
    link='logistic', raw_predictions='output.txt'
)