Here is a law I wish we had for social media. Let's just align incentives.
If you are working on a social media platform, and your total comp is >= 500k, your children must use your platform for the average amount of time that children of that age range use your platform nationwide.
> "Instead of picking the highest-probability token, we can use different selection strategies to balance safety and creativity in the generated text".
Safety is definitely the wrong word here.
Temperature 0 generated text actually has a weird "lack of surprise" character that makes it seem artificial. [1]
> "high-probability texts can be dull or repetitive. Humans use language as a means of communicating information, aiming to do so in a simultaneously efficient and error-minimizing manner; in fact, psycholinguistics research suggests humans choose each word in a string with this subconscious goal in mind."
I'd completely drop the dropout explanation. It's just not part of the modern recipe anymore, AFAICT.
As for the ambitious goal of explaining transformers with a single interactive visualization, I just have a hard time imagining a person is going to newly understand both word embeddings (word2vec blew my mind in 2014) and also gain an understanding of attention.
I am making my own visualizations for a presentation on "Full Bandwidth Transformers"[2] that I am giving tomorrow at the Deep Learning Study Group (SF) (on zoom for the non-locals)[3]. It's not meant to be stand alone/context free, but I'd love some feedback.
> I'd completely drop the dropout explanation. It's just not part of the modern recipe anymore, AFAICT.
Is there any other mechanism replacing it? As far as I understood, while following the LLMs from Scratch book, the dropout is a mechanism to prevent overfitting, and it makes a lot of sense (basically cripple some neuron connections randomly during training such that their weights aren't updated).
Edit: well, I could have searched on the internet, but other people may stumble on your post and you seem quite knowledgeable of the inner workings of these things. :)
Hey, take it easy on the guy. Words are just probabilities in some high-dimensional space; it's a probabalistic selection anyways; he probably just nudged the latent vector a little too hard. Spymarking and watermarking are for safety reasons, you know. There was zero loss of quality there.
> That's an interesting choice. One question I had when looking at the jev copy on their blog is if one "line" in their output looks / attends to other lines. I think not, since they say it's parallel and not autoregressive.
I don't understand the connection between the lack of autoregression and options attending to each other.
Non autoregressive models can attend to all the inputs simultanously.
An autogregressive model can can attend to all the options in the context of each other by simply writing the options out twice. Autoregressive models actually requires this, since one of them will come later, and the earlier prefill inputs can't attend to the later ones.
I haven’t looked into the history, but my understanding from the comments is that it exists to preserve `x == y -> hash(x) == hash(y)` for `x` and `y` with different numeric types (int, float, decimal) – which seems like it should be solvable, but isn’t a constraint I’d want to be working under, so I sympathize.
His inputs are large numbers that don't fit in a standard integer. Bigints. The set inclusion test not only has a hash lookup but an equality test, which will be a bigint comparision rather than integer comparison, and bitint comparison is itself O(n) based on the size of the bignum. And the code that tests each bignum is in the set also _sums_ those bignums, which itself is an O(n) operation based on the size of the bignums being summed.
So he's not testing dict/set performance, he's testing bignum performance, because of the inputs he deliberately chose
Should LLMs be designed to be modular, so that instead of needing access to the whole model, for a given prompt, only a small subset of the model would be used? If knolwedge was sufficiently modularized, most of it could be ignored.
Maybe a hyopthetical model of 5T of indexable weights could be used with only 50 GB of GPU ram, efficiently, because it stays resident in the GPU.
No, not really, current MoE limit the computation, not memory requirements. Router experts are not "sticky" enough to achieve what robrenaud describes - they'd have to be chosen per prompt, or at least per chunk, not per token.
Experts vary per token in MoE, there is maximum flexibility. Good for driving down loss, bad for locality/gpu memory/bandwidth.
If expert selection were more constrained, inference systems could take advantage of it. Keeping experts cached would mean not needing to load them from disk/ram every token.
It's cool that it proves that a bunch of vectorized outputs from an unknown embedder on an unknown dataset is in no way private, because of this ability to reverse engineer the embedder.
I talked to the author at his poster session at neurips and was able to get the gist, though I had read a lot about the platonic representation hypothesis, and this was one of my top 10 favorite papers in the conference.
If you are working on a social media platform, and your total comp is >= 500k, your children must use your platform for the average amount of time that children of that age range use your platform nationwide.
reply