Why do OpenAI's GPT-2 weights beat mine? Part two: the bugfix :: Giles' blog
el.dataset.currentDropdown = '')<br>}">
Giles' blog
Writing the post that I wished I'd found when I started learning whatever it was...
About
Contact
Archives
Categories
Blogroll
July 2026 (6)
June 2026 (7)
May 2026 (2)
April 2026 (11)
March 2026 (3)
February 2026 (4)
January 2026 (4)
December 2025 (1)
November 2025 (3)
October 2025 (9)
September 2025 (3)
August 2025 (5)
July 2025 (1)
June 2025 (2)
May 2025 (3)
April 2025 (2)
March 2025 (7)
February 2025 (10)
January 2025 (6)
December 2024 (7)
September 2024 (1)
August 2024 (2)
July 2024 (2)
May 2024 (2)
April 2024 (2)
February 2024 (2)
April 2023 (1)
March 2023 (2)
September 2022 (1)
February 2022 (1)
November 2021 (1)
March 2021 (1)
February 2021 (2)
August 2019 (1)
November 2018 (1)
May 2017 (1)
December 2016 (1)
April 2016 (1)
August 2015 (1)
December 2014 (1)
August 2014 (1)
March 2014 (1)
December 2013 (1)
October 2013 (3)
September 2013 (4)
August 2013 (2)
July 2013 (1)
June 2013 (1)
February 2013 (1)
October 2012 (1)
June 2012 (1)
May 2012 (1)
April 2012 (1)
February 2012 (1)
October 2011 (1)
June 2011 (1)
May 2011 (1)
April 2011 (1)
March 2011 (1)
February 2011 (1)
January 2011 (1)
December 2010 (3)
November 2010 (1)
October 2010 (1)
September 2010 (1)
August 2010 (1)
July 2010 (1)
May 2010 (3)
April 2010 (1)
March 2010 (2)
February 2010 (3)
January 2010 (4)
December 2009 (2)
November 2009 (5)
October 2009 (2)
September 2009 (2)
August 2009 (3)
July 2009 (1)
May 2009 (1)
April 2009 (1)
March 2009 (5)
February 2009 (5)
January 2009 (5)
December 2008 (3)
November 2008 (7)
October 2008 (4)
September 2008 (2)
August 2008 (1)
July 2008 (1)
June 2008 (1)
May 2008 (1)
April 2008 (1)
January 2008 (4)
December 2007 (3)
March 2007 (3)
February 2007 (1)
January 2007 (2)
December 2006 (4)
November 2006 (18)
AI (93)
TIL deep dives (77)
Python (73)
LLM from scratch (48)
Resolver One (34)
PyTorch (21)
TIL (21)
Blogkeeping (18)
PythonAnywhere (17)
Linux (16)
Startups (15)
Gadgets (13)
Hugging Face (13)
NSLU2 offsite backup project (13)
Funny (11)
Musings (11)
Finance (10)
Fine-tuning LLMs (10)
C (9)
Personal (8)
Robotics (8)
Website design (8)
JAX (7)
3D (5)
Quick links (5)
Rants (5)
Cryptography (4)
JavaScript (4)
Music (4)
Oddities (4)
Talks (4)
Dirigible (3)
Eee (3)
Memes (3)
Politics (3)
Django (2)
GPT-2 mysteries (2)
GPU Computing (2)
LaTeX (2)
MathML (2)
Microprojects (2)
OLPC XO (2)
Retro Language Models (2)
Space (2)
VoIP (2)
Copyright (1)
Golang (1)
poppy the training box (1)
Raspberry Pi (1)
Software development tools (1)
Agile Abstractions
Astral Codex Ten
:: (Bloggable a) => a -> IO ()
David Friedman's Substack
Econ & Energy
Entrepreneurial Geekiness
For some value of "Magic"
Hackaday
kaleidic.ai newsletter
Knowing.NET
Language Log
Millennium Hand
ntoll.org
Obey the Testing Goat!
PK
PythonAnywhere News
Simon Willison's Weblog
Societive
Software Deviser
Some opinions, held with varying degrees of certainty
tartley.com
Why do OpenAI's GPT-2 weights beat mine? Part two: the bugfix
Posted on 30 July 2026
in
GPT-2 mysteries,
AI,
Python
I'm digging into why my GPT-2 style models score worse on an instruction-following eval than OpenAI's original weights;<br>I gave the details in this post.
While I was writing up the results of my first experiment into possible causes,<br>I ran the post past ChatGPT -- I always use an "editorial board" of AIs to check my<br>posts for flow, style, and any technical errors (though all writing is always mine).<br>It took a look at the eval code that I was running, and highlighted a bug.
Luckily, it doesn't change the important results -- OpenAI's models continue to be<br>better than mine at instruction-following. But it was<br>enough to change the baseline numbers, re-ordering how well my own models did.<br>So I fixed it and regenerated the baseline<br>so that future experiments are based on solid ground.
The bug
The eval takes a model, and trains it over multiple epochs on a split of a subset of the Alpaca<br>instruction-following dataset. At the end of each epoch, it evaluates<br>the resulting model against a held-back validation split; if the eval loss starts<br>rising, it bails out. Finally, it runs a test split of the dataset through the resulting<br>model, and saves the result.
Once I've run it for a bunch of different models, I use an LLM-as-a-judge script<br>to get GPT 5.5 to score results -- for each question-answering result, it sees all<br>of the responses for all of the models in the same prompt, shuffled in order each time,<br>to try to make it judge models against each other as consistently as possible.
Now, the idea was that the generation of the test split answers would use the model from the epoch prior<br>to the rising-loss one. So I had code like this:
for epoch in range(100):<br>model.train()<br>for...