> [in 2020/2021] the dominance of autoregression was not as well-established as it is today: GPT-3 had turned some heads, but the ‘ChatGPT moment’ wouldn’t come until late 2022
I disagree with this. Decoders were absolutely dominant in 2020 for chat. GPT2 was considered too dangerous to release, and I remember scrambling to get on the GPT3 waitlist. It worked.
(The only exception I will make is encoder-decoder models which now are often done by decoder-only.)
But what made it go mainstream was RL. RLHF at first, then other improvements like DPO that were less of a pain in the ass to set up. Adding diffusion on top of that would be an even bigger pain in the ass.
Before ChatGPT there really wasn’t much of a concept of pre-training and post-training. It was all pre-training. Post training was what made the bots conversational and not just “continuing the thing you wrote to them”.
So in short, diffusion never took off because it was just a more complicated way to generate tokens, and the real problem was getting tokens in the right distribution.
> I disagree with this. Decoders were absolutely dominant in 2020 for chat. GPT2 was considered too dangerous to release, and I remember scrambling to get on the GPT3 waitlist. It worked.
He's not talking about decoders, he's talking about auto-regression. Before ChatGPT, the dominant paradigm was fine-tuning BERT-like models.
> Before ChatGPT there really wasn’t much of a concept of pre-training and post-training.
Again, people spend years just post-training BERTs in various ways.
I agree that this should be something that researchers reflect on. GPT-2 is one of the primary models to research on nowadays, and many recent developments have come from studying it as a test bench.
Imagine if CRISPR was considered "too dangerous" to publish because of the potential ethical ramifications, and that only a special few should be aware. It is utter self-righteousness, and it is shameful behaviour. The world cannot adjust itself to what it cannot see, so you risk greater catastrophe by keeping it secret.
The open dissemination of knowledge at every increment is the only way for society to truly deal with what is to come.
My impression at the time was they were perhaps overly cautious but this was a bunch of researchers who wanted to self-regulate. Anthropic didn’t exist, deepmind was also much less product-focused and relatively cautious. Chinese models weren’t really a factor either.
Government regulation was really not in the picture either in 2020 or 2021 tbh. The government was still trying to beat a pandemic. Some of the Biden admin eventually wanted to but it wasn’t very serious.
Parent post is saying there was no community of peer models, just some people figuring it out as they went and with lots of slack to go slow if they wanted
There were other players, china included, but it wasn't relevant until LLMs became profitable? China is being singled out as usual because it's the antithesis of the us
Well, no. Without controls, a language model can drive sensitive individuals to violence or suicide. The idea of releasing a frontier model without RL is frightening based on what we have learned.
> The idea of releasing a frontier model without RL is frightening
In case you were not aware, strong base models (no post training at all) have been available for quite some time now. Including ones that eclipse “scary” frontier models from even a year ago.
Words out of a magic box on a computer can’t make you kill yourself. Perhaps the people that would use that as encouragement are already mentally ill enough that it really doesn’t matter what the trigger is?
On a more callus but fully serious note, evolution starts out physical, that the species that don’t eat and breed as well, die. What happens when you remove that? When life is so safe that you basically wont stave, get eaten, catch a disease, don’t really need to compete all that hard for scarce resources, etc? Do you think evolution just stops? Or perhaps does a social or mental evolution become the predominant differentiator for successful reproduction over time?
I would love to see models that can think at different rates and also output a thinking scratchpad alongside output text instead of before all output.
Right now models need to rely on less legible compressed CoT to get high intelligence per token/step, but with diffusion they would just need to output more tokens per step instead.
There's a cool research project https://github.com/exoharness/exo that is designed specifically as a self-modifying harness, so architecturally it separates things in a way that makes it a lot harder for the agent to break itself when modifying itself. It looks pretty neat. Saw the author do an interview on a podcast explaining it.
> [in 2020/2021] the dominance of autoregression was not as well-established as it is today: GPT-3 had turned some heads, but the ‘ChatGPT moment’ wouldn’t come until late 2022
I disagree with this. Decoders were absolutely dominant in 2020 for chat. GPT2 was considered too dangerous to release, and I remember scrambling to get on the GPT3 waitlist. It worked.
(The only exception I will make is encoder-decoder models which now are often done by decoder-only.)
But what made it go mainstream was RL. RLHF at first, then other improvements like DPO that were less of a pain in the ass to set up. Adding diffusion on top of that would be an even bigger pain in the ass.
Before ChatGPT there really wasn’t much of a concept of pre-training and post-training. It was all pre-training. Post training was what made the bots conversational and not just “continuing the thing you wrote to them”.
So in short, diffusion never took off because it was just a more complicated way to generate tokens, and the real problem was getting tokens in the right distribution.
> I disagree with this. Decoders were absolutely dominant in 2020 for chat. GPT2 was considered too dangerous to release, and I remember scrambling to get on the GPT3 waitlist. It worked.
He's not talking about decoders, he's talking about auto-regression. Before ChatGPT, the dominant paradigm was fine-tuning BERT-like models.
> Before ChatGPT there really wasn’t much of a concept of pre-training and post-training.
Again, people spend years just post-training BERTs in various ways.
> GPT2 was considered too dangerous to release
This is how ridiculous this industry is. Regulation-seeking panic over nothing. Drama in search of a moat.
Everything is "too dangerous". GPT2 is going to invent a time machine and break crypto and genetically engineer super rabies.
They sell knives, guns, combustible materials, and multi-ton heavy machinery in stores. That's what's actually dangerous.
I agree that this should be something that researchers reflect on. GPT-2 is one of the primary models to research on nowadays, and many recent developments have come from studying it as a test bench.
Imagine if CRISPR was considered "too dangerous" to publish because of the potential ethical ramifications, and that only a special few should be aware. It is utter self-righteousness, and it is shameful behaviour. The world cannot adjust itself to what it cannot see, so you risk greater catastrophe by keeping it secret.
The open dissemination of knowledge at every increment is the only way for society to truly deal with what is to come.
What are your thoughts on the HuggingFace incident?
"Drama in search of a moat" is hard to beat.
Optimal next token generation disguised as something more.
My impression at the time was they were perhaps overly cautious but this was a bunch of researchers who wanted to self-regulate. Anthropic didn’t exist, deepmind was also much less product-focused and relatively cautious. Chinese models weren’t really a factor either.
Government regulation was really not in the picture either in 2020 or 2021 tbh. The government was still trying to beat a pandemic. Some of the Biden admin eventually wanted to but it wasn’t very serious.
why are chinese models a consideration at all?
Parent post is saying there was no community of peer models, just some people figuring it out as they went and with lots of slack to go slow if they wanted
Oh it's the typical china race thing, we can't let the antithesis of the us be better than the us etc. American exceptionalism
The 'china' part is irrelevant. You're addicted to being mad. Parent post is just saying there were no other players
There were other players, china included, but it wasn't relevant until LLMs became profitable? China is being singled out as usual because it's the antithesis of the us
Well, no. Without controls, a language model can drive sensitive individuals to violence or suicide. The idea of releasing a frontier model without RL is frightening based on what we have learned.
> The idea of releasing a frontier model without RL is frightening
In case you were not aware, strong base models (no post training at all) have been available for quite some time now. Including ones that eclipse “scary” frontier models from even a year ago.
So…
Words out of a magic box on a computer can’t make you kill yourself. Perhaps the people that would use that as encouragement are already mentally ill enough that it really doesn’t matter what the trigger is?
On a more callus but fully serious note, evolution starts out physical, that the species that don’t eat and breed as well, die. What happens when you remove that? When life is so safe that you basically wont stave, get eaten, catch a disease, don’t really need to compete all that hard for scarce resources, etc? Do you think evolution just stops? Or perhaps does a social or mental evolution become the predominant differentiator for successful reproduction over time?
I would love to see models that can think at different rates and also output a thinking scratchpad alongside output text instead of before all output.
Right now models need to rely on less legible compressed CoT to get high intelligence per token/step, but with diffusion they would just need to output more tokens per step instead.
It’s refreshing to read something not AI generated.
I feel like there is still low hanging fruit on the auto regressive LLMs; the encoder
CDLMs sound promising for smoother, more coherent text generation. Excited to see how they tackle the token-level discontinuities.
Diffusion models for language felt inevitable. Imagine the creative potential once these mature beyond current limits.
A great read - as with all of Sander's diffusion posts.
I wonder if we’ll get something like CDLMs for automated harness engineering, sort of piloting the LLM underneath.
how do tools like hermes do this? does it just review sessions and rewrite markdown files?
also haven't read too deep into the deepseek agent harness but the math in there was really cool. it sounded promising, at least.
There's a cool research project https://github.com/exoharness/exo that is designed specifically as a self-modifying harness, so architecturally it separates things in a way that makes it a lot harder for the agent to break itself when modifying itself. It looks pretty neat. Saw the author do an interview on a podcast explaining it.
> does it just review sessions and rewrite markdown files
Yes, same with openclaw etc. Some might have plugins to integrate with graph or vector databases besides only markdown files.
"Attention is all you need" should be renamed into "Attention is sufficient but not necessary".
It’s the opposite: attention is necessary but not sufficient.
Just about anyone building a large diffusion model is relying on transformers
A quick look at the continuous diffusion models linked in the post shows lots of transformer models still