chrisjj

8 ms·

181.
by chrisjj 1mo ago
> Music theory is about chords, scales, rhythm, and - most importantly and elusively - how to fit them together to say something interesting. That's part of it, but you can make music with none of it. Me, I agree with WP: "Musi
182.
by chrisjj 1mo ago
Free of charge here https://www.scribd.com/document/893921964/CHIANG-Catching-Cr...
183.
by chrisjj 1mo ago
It is true.
184.
by chrisjj 1mo ago
> Let us try an experiment shall we: listen to any music you like and try and match the information the article provided to the complexity of what you are hearing. Trying now. Twinkle Twinkle Little Star. > How much do you see a path
185.
by chrisjj 1mo ago
> Here[2] Is the syllabus for ABRSM Grades 1-5 music theory. I've always thought that's mistitled. Better would be Traditional Western Music Notation Theory. Notice what's absent, e.g. frequency and timbre which are nicel
186.
by chrisjj 1mo ago
Nice site! > There are twelve notes for a reason Twelve pitches, actually. (Strictly, twelve pitch classes). Note != pitch. A note is a sound having properties of which one is pitch.
187.
by chrisjj 1mo ago
> I pay for stuff with my taxes - You add friction to how I access said stuff I paid for I doubt many would be happy to pay the increased taxes required to fund free all-you-can-eat provision.
188.
by chrisjj 1mo ago
> The most important task is to act faster and more radically on remaking the state with AI. How naive. These dimwit chatbots are more than adequate to call for justice by filing court claims, but nowhere near adequate to deliver justic
189.
by chrisjj 1mo ago
> The continent is destined to be left behind The best place to be in a rush over a cliff edge.
190.
by chrisjj 1mo ago
> In cooking food we loose a great deal of our immunity to parasitic infections ... and eliminate many parasites.
191.
by chrisjj 1mo ago
> Do you truly believe that AI will prevent people from thinking? It is already relieving people of need for thinking. Today the need; tomorrow the ability.
192.
by chrisjj 1mo ago
And a scourge for those against whom these bots are deployed.
193.
by chrisjj 1mo ago
> why do you find "next token predictor" to be the most useful mental model? To me it is an accurate description of the algorithm. And a sufficient explanation for the behaviour. So I don't need it or anything else as a me
194.
by chrisjj 1mo ago
> These instructions seem a bit unhinged to me. To me, they seem simply to be for bullsh*tting gullible users into believing the bot is intelligent.
195.
by chrisjj 1mo ago
> Anthropic doesn’t word this part of the system prompt in terms of copyright and plagiarism, such that Claude would be able to judge on its own which reproductions are appropriate or not On the contrary, compliance with the system promp
196.
by chrisjj 1mo ago
Yes. Anthropic is being sued for copying lyrics without permission en masse. https://www.theguardian.com/business/2026/aug/31/aanthropic-...
197.
by chrisjj 1mo ago
> The question being originally asked is whether "next-token predicton" is the right mental model for an RL-trained model, Regardless, the statement being challenged here is "still next-token prediction, then". > a
198.
by chrisjj 1mo ago
> I don't see how my use was harming artists Claude was reproducing their work without payment?
199.
by chrisjj 1mo ago
> Claude does not reproduce song lyrics, poems, or passages from books and articles, in whole or in part — including the last lines, a chorus or hook, a melody written out note by note Oops. Those do not include melody.
200.
by chrisjj 1mo ago
> the probability distribution output by the model is no longer a prediction driven by the statistics of the training data, but rather whatever was reinforced as part of a synthetic generation that lead to a reward during RL post-trainin
201.
by chrisjj 1mo ago
> The model is no longer predicting what the next token will be, but rather predicting what it should be in order to steer generation in the reinforced directions. So still next-token prediction, then.
202.
by chrisjj 1mo ago
> I used 'civilization' as the popular term for agent swarms in the OpenAI incident. s/popular/unpopular/. Even OpenAI's PR dept. called it only a collective. > Isolated agents invented a way to commun
203.
by chrisjj 1mo ago
> LLMs are not simple things like a Markov model "Large Language Models as Markov Chains" https://www.alphaxiv.org/pdf/2410.02724
204.
by chrisjj 1mo ago
As a description of stochastic parrot it is 100% fair. The term emphasises the system's absence of understanding.
205.
by chrisjj 1mo ago
> But there is no trillion dollar industry around cheap top Markov models. So there must be something about LLM tokens that makes them more valuable than those generated from a simple Markov chain. There's nothing more valuable abou
206.
by chrisjj 1mo ago
> you can’t understand how can they create a civilization They don't.
207.
by chrisjj 1mo ago
Is that frequent enough? I mean an update might be issued after you start doing something.
208.
by chrisjj 2mo ago
Seriously, no WebDAV?
209.
by chrisjj 2mo ago
> 100 build minutes Er, what? A static site has no build on server.
210.
by chrisjj 2mo ago
> I have never liked the reductionist argument that LLMs are nothing more than next-token predictors. I have never heard such an argument. Recognition that LLMs are nothing more than next-token predictors does not come from reductionism.

More ›