upvote
These aren’t raw base models they are the result of a ton of RLHF and various adjustments.

Bitter lesson wildly overstated in this context.

reply
Not sure where I implied they are "raw base models" and not sure what "various adjustments" means here or how "ton of RLHF" contradicts anything. If we look at research for open source models, "adjustments" usually come in the form of efficiency gains which directly contributes to the ability to scale or synthetic data pipelines to increase the dataset and increasing the context window.
reply
rlhf = reinforcement learning from human feedback

(had to look it up)

reply
I think it's more RLVR (reinforcement learning from verified rewards). The RLHF is just to align models to human preferences, meaning to behave nice.
reply
What makes you say that
reply
More RLHF is in fact scaling.
reply
Yes, but not in the “dump another chunk of all written language in the bucket and stir”-sense which is what bitter lesson became synonymous with.

That may not be the intent of the original article, but over the past few years that’s what the phrase turned into.

reply
GPT-6 is supposed to be using a much larger base model that just finished pretraining so the "dump another chunk of all written language" approach is still going strong.
reply
Modern pretraining also consists of expensive human-led specialized task creation and grading loops. Synthetic generation and distillation from previous models is another input for training. I wonder how much new text contributes beyond keeping knowledge up-to-date.
reply
The measure is to see if the results scale, not just the rumored attempts at building such a model, as o3 taught us.
reply
Where are they getting new data from at this point? Didn't they already read the entire internet?
reply
Custom sets and mining their own users, as every lab does
reply
The bitter lesson just means “compute scaling beats hand-tuned architectures in the long run”.

As GP said. More RLHF is in fact the bitter lesson.

reply
deleted
reply
Not just RLHF but also RLVR, and isn't that the litter lesson though?

My sense of the Sutton Dwarkesh interview was that he was calling out that he didn't mean just longer datasets, but rather learning through exploration and that's exactly RL.

reply
They just need more contact with reality. That's what RL is right? Contact with narrow subsets of reality.
reply
RLHF is an increasingly small part of training though? From what I understand most of the capability gain is in RLVR
reply
Nah, the last few generations have more RLVR in the data mix. Which is more CPU intensive and very much amenable to the bitter lesson as you can reduce the loss by doing more rollouts in your tool environment.
reply
The scaling with reasoning models is more and more with things like verifiable rewards (coding and math), in line with bitter lesson and also Sutton invented lots of modern RL.
reply
While I think this is true, remember as we get more efficient we just decide to scale even bigger. So more GPUs, and more efficient.

I agree with the sibling comment, effiency is probably the more important component at this point. We are hitting not just a practical engineering roadblock for scaling with current technology, I think we have definitely hit a financial and logistical roadblock for up scaling with the number of GPUs (on an immediate basis)

reply
There goes my plan to buy a PC for the next decade
reply
The whole of knowledge work is being automated. We've barely begun to see the GPU build out. This is just the start.

I'd imagine they're going to 10x this, maybe 100x this.

reply
yeah, spacex's planned million satellite datacenter-in-space constellation doesn't seem as absurd.
reply
Sam Altman was saying we'll eventually need to build a dyson sphere. We'll see how far human society can withstand this kind of escalation. I'm guessing one more year before there are major fractures in the basic economic/social/political relationship that get so bad it actually prevents more buildout.
reply
I actually think we're in a strange situation with AI compute.

Right now, we have models that are statistical models of language, with a world model and reasoning "falling out" of a lot of effort.

It's like we've made something that's a little bit intelligent, and now we're trying to amplify that trick to create something that's quite intelligent. And - don't get me wrong - it works.

But it's also super, super inefficient. We're having machines "think out loud" to compensate for the quality of their thought processes. We elongate the path to make up for the progress made on a given step.

I tink there's probably a much smarter way of doing things that will require qualitative architectural (and quite possibly hardware) innovations. Right now we're on the path to a Dyson sphere: that's probably not going to be necessary once we figure out a smarter way to think.

reply
Not always, in some cases, changing to a higher reasoning makes the AI doubt itself too much, and skip over the correct answer by overcomplicating the problem and polluting the context.

It would be nice to see on which categories of problems the extra thinking makes it better and on which it makes it worse.

reply
This shows up in OpenAI's graphs on their announcement page. There is a peak performance datapoint in the graphs past which (to the right on the graph indicating more resources spent) peformance declines. And it's on every graph on that page!
reply
And in my tests, that point of "overthinking" depends on the problem's complexity, so it's not necessarily that using "xhigh" is always bad or good.
reply
I think I have this problem but with my human brain.
reply
Kind of refreshing though that the "throw more processing at it" scaling we saw in the 90s has returned in a different way. For a while we were really bottlenecked in our advances by relatively low levels of parallelism (most software used by your average user doesn't scale cleanly with more than a few threads).
reply
I mean, theoretically you can solve every finitary problem with a brute force solution...

Richard Sutton specifically states that the search has to be smart. We know that the brain uses recurrent connections and is shallow. I think a lot more money has to go into architecture. Feed Forward transformers can only scale so far

reply
We’re definitely going to need a lot of Gpu’s
reply
Or a new model. The human brain does far more with far less.
reply
Yes an no current models can read and output much more faster with differing quality tho.
reply
Or a lot better efficiency.
reply
I said a few months ago, "man, Opus is great, but sometimes when talking with it I have the feeling like, this thing should be about 10 times bigger."

When Mythos was announced after that, I was pleasantly surprised to hear about it. But when it turned out to be only two times bigger, I was a little disappointed!

(I am even more disappointed with the safety filters, but that's kind of a separate discussion... "Fortunately" I find that I can usually edit my prompt by single character and get through...)

reply
What do you mean bigger? Bigger functional context window?
reply
> Dramatic difference

Isn't this just the difference between getting 0 right and getting 1 right?

reply
And a lot more electricity to power them.
reply
And dozens of data centers in every state so tokens are dirt cheap.
reply
> We are probably going to need a lot more GPUs.

Or a breakthrough in algorithms etc.

The human brain, heck all bio brains, are proof that you don't need a lot of power or size for intelligence.

reply
The human brain has 80 billion neurons and a 100 trillion synapses. I think you're underselling the processing power of that warm chunk of meat.

The real message of the last 15 years has actually been the opposite: if you throw enough processing power at it, intelligence emerges.

reply
Moreover we've known for quite a while now that glial cells also participate in cognition and moderate learning (e.g.: [1]). When you take those connections into account the numbers get really staggering. 85 billion glial cells with trillions of protein channels facilitating communication between the glial syncytium [2].

[1] https://www.sciencedirect.com/science/article/pii/S193459091... [2] https://pmc.ncbi.nlm.nih.gov/articles/PMC5063692/

reply
It’s still 20W. We have living proof what is possible within 20W. The message has always been clear - try to get silicon computer to be as power efficient as the brain is as it is obviously possible.
reply
I keep thinking about the cyborg chicken mechs in Metal Gear Solid 4. If stuff like Neuralink gets advanced enough, we might find that cutting up animals and sticking electrodes in their brains is the best way to make robots.

Yeah, people might object, but it can be argued that we are already subjecting scores of animals to horrors beyond comprehension just to get a bucket of chicken wings. And even if we manage to get silicon to do what brains do, it will likely cost 1000x as much and consume 1000x the power like you said.

It's hell of an economic incentive.

reply
The real question is not how many "weights" the human brain has (neurons+synapses may or may not translate into "weights", and brain might be also inefficient for what it is), but rather how much evolutionary and social "compute" was necessary to pack everything into that capacity.
reply
I think you're helping GPs point: there is a lot of efficiency gains to be made to match the processing power of the brain, given it's size and power draw.
reply
deleted
reply
20 watts for inference AND training!
reply
For intelligence, I expect the next breakthrough to be colocation of memory and compute in the same chip. And we'll need much more of this memory, probably a few petabytes.
reply
This isn’t really how it works anymore. Agents rely heavily on tool use and the agentic harness to perform tasks. Pre-training is no longer very effective.
reply
I thought models werent allowed tools on arc-agi?
reply