BurakaKrishna

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 and experience preferred (super rare at this point).

My own notes from a few months back.

Background

• Survey of LLMS
• Self-attention and transformer networks

BurakaKrishna

This worked on 14/May/23. The instructions will probably require updating in the future.

llama is a text prediction model similar to GPT-2, and the version of GPT-3 that has not been fine tuned yet. It is also possible to run fine tuned versions (like alpaca or vicuna with this. I think. Those versions are more focused on answering questions)

Note: I have been told that this does not support multiple GPUs. It can only use a single GPU.

It is possible to run LLama 13B with a 6GB graphics card now! (e.g. a RTX 2060). Thanks to the amazing work involved in llama.cpp. The latest change is CUDA/cuBLAS which allows you pick an arbitrary number of the transformer layers to be run on the GPU. This is perfect for low VRAM.

• Clone llama.cpp from git, I am on commit 08737ef720f0510c7ec2aa84d7f70c691073c35d.

BurakaKrishna

import numpy as np
values = df[column].tolist()
n = len(df[column].tolist())
counts = np.zeros(n, dtype='int64')    
for i in range(n):
    v = values[i]
    if len(v):
        counts[i] += len(v)
    else:
        # empty list-like, use a nan marker

BurakaKrishna

#Understanding neural networks
#Random approach
def forward_multiply_gate(x,y):
	return x*y

x = -2
y = 3

tweak_amount = 0.01
import random

BurakaKrishna

"""
Usage: python remove_output.py notebook.ipynb [ > without_output.ipynb ]
Modified from remove_output by Minrk

"""
import sys
import io
import os
from IPython.nbformat.current import read, write

BurakaKrishna

import string
import math

tokenize = lambda doc: doc.lower().split(" ")

document_0 = "China has a strong economy that is growing at a rapid pace. However politically it differs greatly from the US Economy."
document_1 = "At last, China seems serious about confronting an endemic problem: domestic violence and corruption."
document_2 = "Japan's prime minister, Shinzo Abe, is working towards healing the economic turmoil in his own country for his view on the future of his people."
document_3 = "Vladimir Putin is working hard to fix the economy in Russia as the Ruble has tumbled."
document_4 = "What's the future of Abenomics? We asked Shinzo Abe for his views"

BurakaKrishna

Movies Recommendation:

• MovieLens - Movie Recommendation Data Sets http://www.grouplens.org/node/73
• Yahoo! - Movie, Music, and Images Ratings Data Sets http://webscope.sandbox.yahoo.com/catalog.php?datatype=r
• Jester - Movie Ratings Data Sets (Collaborative Filtering Dataset) http://www.ieor.berkeley.edu/~goldberg/jester-data/
• Cornell University - Movie-review data for use in sentiment-analysis experiments http://www.cs.cornell.edu/people/pabo/movie-review-data/

Music Recommendation:

• Last.fm - Music Recommendation Data Sets http://www.dtic.upf.edu/~ocelma/MusicRecommendationDataset/index.html

BurakaKrishna

# post_loc.txt contains the json you want to post
# -p means to POST it
# -H adds an Auth header (could be Basic or Token)
# -T sets the Content-Type
# -c is concurrent clients
# -n is the number of requests to run in the test

ab -p post_loc.txt -T application/json -H 'Authorization: Token abcd1234' -c 10 -n 2000 http://example.com/api/v1/locations/

BurakaKrishna

The Best of the Best Practices (BOBP) Guide for Python

A "Best of the Best Practices" (BOBP) guide to developing in Python.

In General

Values

• "Build tools for others that you want to be built for you." - Kenneth Reitz
• "Simplicity is alway better than functionality." - Pieter Hintjens