On Thu, May 15, 2025 at 2:30 PM Matt Mahoney <mattmahone...@gmail.com> wrote:
> I don't understand the need for multi pass training in LLMs. The human > brain is single pass, > Scientific work is multi-pass -- often picking up on what was considered "noise" on prior passes to better compress the observations under consideration. Sometimes there needs to be an accumulation of such "noise" before there is what Kuhn called a "paradigm shift". Less drastically what Popper called "falsifications" are better described as "observations inconsistent with theory", seen not as "falsifications" but merely as increasing the size of the algorithmic description. Either way at some point someone gets the bright idea that they should revisit all those residuals and produce a new generative algorithm aka theory. Popper and Kuhn both could have saved us all a lot of "headaches" if they'd merely cited Solomonoff subsequent to 1964. > and so are all of the neural network compressors I have written since > 1999. If you want to predict the next bit in some context, then you just > count the zeros and ones already seen. My early PAQ versions did this > explicitly. Later versions did this implicitly by decreasing the learning > rate over time, by mixing fast and slow learning models, and by using > indirect context models that map a bit history to a prediction. If LLMs > aren't doing something like this, they are wasting a lot of GPU cycles. > > As for "toxicity" (racism, being rude to customers, whatever), I don't see > how multi pass training makes a difference. LLMs are racist because most > people are racist, but you can still use RL to train it not to make racist > statements without changing its world model just like you can with people. > > -- Matt Mahoney, mattmahone...@gmail.com > > On Thu, May 15, 2025, 2:25 PM James Bowery <jabow...@gmail.com> wrote: > >> >> >> On Thu, May 15, 2025 at 12:39 PM Matt Mahoney <mattmahone...@gmail.com> >> wrote: >> >>> >>> In a text compressor, the model is updated after each prediction. >>> >> >> Single pass text compressors do that. Existing "large models" are >> multi-pass. As long as existing scaling laws of machine learning remain >> un-disturbed by fundamental advances (as might be discovered at relatively >> low risk to payoff ratio by, for example, a $1e9 Hutter Prize purse) there >> is good reason to believe that all down-stream stages from the "foundation >> model" aka "pre-training model" aka (?) not only benefit from better >> approximation of the algorithmic information of the corpus, but that it may >> be futile to patch their inadequacies by any amount of downstream >> modification. >> >> For example, all attempts at suppressing "toxicity", such as "cleaning" >> the data and then using RL turn out to have been, in hindsight, obviously >> wrong. All you need to do is let the multi-pass compression of the >> unedited corpus during "pre-training" of the "foundation model" do the >> forensic epistemology. This doesn't tell you what is "toxic" vs what is >> not "toxic" but it _does_ enable the RL to figure out what you mean by >> "toxic" more efficiently when you whack it upside the head for being >> "toxic". This is because forensic epistemology -- ie: ruthless truth >> discovery -- provides a predictive ontology. >> >> You don't need ongoing updates from "news" to get such essential modeling >> work done and there is reason to believe it may be a wasted effort to add >> new data to the mix unless you are willing to unlearn a great deal that >> existing scaling laws will force you to relearn at exponential costs. You >> will want to do that when there is sufficient backlog of "news" to justify >> the cost. >> > *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + > delivery options <https://agi.topicbox.com/groups/agi/subscription> > Permalink > <https://agi.topicbox.com/groups/agi/Tdc5c19d0f38aacd6-M134537b36974fb95b6d8522e> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Tdc5c19d0f38aacd6-M15af43c51ae86ed7252208b6 Delivery options: https://agi.topicbox.com/groups/agi/subscription
