Why do OpenAI's GPT-2 weights beat mine?

gpjt1 pts0 comments

Why do OpenAI's GPT-2 weights beat mine? :: 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 (5)

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 (92)

TIL deep dives (77)

Python (72)

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)

GPU Computing (2)

LaTeX (2)

MathML (2)

Microprojects (2)

OLPC XO (2)

Retro Language Models (2)

Space (2)

VoIP (2)

Copyright (1)

GPT-2 mysteries (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?

Posted on 29 July 2026

in

GPT-2 mysteries,

AI

When I finished my project training an LLM from scratch, I was<br>left with a minor mystery. Why were my models worse at instruction-following than the original OpenAI<br>GPT-2 small weights?

I had an evaluation that I was running,<br>based on the instruction fine-tuning code in chapter 7 of<br>"Build a Large Language Model (from Scratch)".<br>The process was to train a model on samples from the Alpaca<br>instruction-following dataset until validation loss started rising, to use that instruction fine-tuned model to generate<br>completions to a held-back test set, and then to use an LLM to compare the<br>results from various different models. The details are here;<br>let's call it the IFT eval.

OpenAI's original weights for GPT-2 small consistently beat my own<br>models, even when mine got better results than theirs on a more technical<br>evaluation, where I just measured the cross entropy loss for each model on a<br>held-back set of test sequences. This surprised me; I would have expected<br>a reasonably close correlation between the two evals -- that better test loss would imply<br>better instruction-following.

I have a couple of thoughts about why this might be, and given that I recently<br>set up poppy, my dedicated LLM training box<br>I decided to inaugurate her with an experiment to test one of them; further<br>experiments will come in time -- though I don't think this will be a focus for the blog. More<br>of a running theme, with occasional posts until either I solve the mystery, or<br>give up in despair...

In this post I'll give a bit more detail about the nature of the problem, and<br>list some of the things I've been thinking might be the cause. In the next post, I'll give the results<br>of the first experiment.

More on the mystery

Let's take a look at the results...

february april december march august july

Related Articles