Duverger PETGA

Fondateur de FaStyle | Architecte de Solutions AWS | Développeur React | Apprenti des Systèmes Distribués et de l'Intelligence Artificielle.

[Duverger PETGA @epegase]

J'ai décidé de préparer un cours sur les LLMs (Large Language Models) et j'essaie de boucler le budget du cours a travers un crowdfunding ciblé sur les acteurs de l'IA. Je vous donne les résultats de cette experience avant la fin de la semaine.

[Nassim Nicholas Taleb @nntaleb]

Egypt-Babylone played an even larger role in science, repackaged as Greek hence "Western".
Historians mistake (implicitly, not necessarily explicitly) "no evidence of earlier" for "evidence of earliest".

[Duverger PETGA @epegase]

Restoration of access to Z-Library and the cessation of illegal criminal prosecution - Signez la pétition ! https://t.co/Y4g7YXVh6D via @Change

[Harry Crane @HarryDCrane]

The defining characteristic of academia:

Big egos, little stakes

[Nassim Nicholas Taleb @nntaleb]

6/ Some misinterpretations of the idea.
A- Studying the anatomy of economic life is important: national accounting, money creation, balance of payment, money flow, institutional finance, trade, etc. But that's NOT what academic economists do. They theorize, often with bad math.

B- That someone like Warren Buffett, Elon Musk,... studied economics is a fake argument. that's not what they do in their function. I know plenty of gold traders who were dentists: you don't study dentistry to trade gold.

[DHH @dhh]

Writebook is our love letter to book-length writing for the internet. Made on the same process as Campfire, code is included, but this time, the software is entirely FREE! Will be publishing the Omakub Manual on Writebook shortly as well. Enjoy!

[Ole Peters @ole_b_peters]

Can I just say again that it's totally stupid to judge how rational people are by giving them probability problems to solve?

Conjunction fallacy... come on!
Probability cannot even be experienced. A fair coin is a mathematical construct. No real coin is knowably fair or biased.

[PrimerYang @yangzhizheng1]

Shocked! Llama3-V project from a Stanford team plagiarized a lot from MiniCPM-Llama3-V 2.5!
its code is a reformatting of MiniCPM-Llama3-V 2.5, and the model's behavior is highly similar to a noised version of MiniCPM-Llama3-V 2.5 checkpoint.
Evidence: https://t.co/L65B6USzJv

[Nassim Nicholas Taleb @nntaleb]

It's the GMO story all over again: shills demonizing processed foods skeptics.

‘Deny, denounce, delay’: the battle over the risk of ultra-processed foods https://t.co/RGsYrNFd7B via @ft

[Nassim Nicholas Taleb @nntaleb]

An academic talking abt freedom is like a nun lecturing on sex.

"Free like an ancient Greek", by a member of "Académie", the LEAST free of all human beings.

As we saw w/Columbia's Shafiq, to be free you need to not depend on the establishment, institutions, and "academie".

[Duverger PETGA @epegase]

From @asmartbear:

Even at wildly successful startups, the first few years are gut-wrenching, uncertain, on the brink of collapse, where pessimism is realism, and yet optimism is required.

https://t.co/DTAdJQBItI

[Nassim Nicholas Taleb @nntaleb]

A severe flaw in Ricardo's model used by idiotic economists:

By sending your manufacturing to China, you also send your accidental DISCOVERY to China.

See #Antifragile : science is more likely to come from technology (and business) than the reverse.

[Paul Biggar 🇵🇸🇮🇪 @paulbiggar]

https://t.co/4nfsQ7NGNX

[Paul Biggar 🇵🇸🇮🇪 @paulbiggar]

Lavender uses WhatsApp messages as input to their AI - where do they get this from? Surely Meta should not be giving group membership to a military that is bombing their users and their entire families, without oversight?

[Duverger PETGA @epegase]

La Lettre de l'Intelligence Artificielle N°12, by @epegase https://t.co/EfQeUsqSEV

[swyx @swyx]

This is still the best* introduction to GPTs I have ever read.

https://t.co/WNVKnQVMtn

Fills in all the gaps between Attention is All You Need right up to coding every single line of a working GPT2 implementation yourself. Even ends with explanations of finetuning for classification, summarization, instruction tuning, and PEFT. Bravo @jaykmody and all who take the time to write these things down in fully commented python rather than cryptic latex.

*best WRITTEN introduction. we all know who has the best video intro to GPTs…

[Nassim Nicholas Taleb @nntaleb]

In finance, people pay money for instructions on "how to get rich and retire" to people who are probably not rich and certainly not retired.

[Nassim Nicholas Taleb @nntaleb]

The extermination of people deemed vermin (after dispossession), that is, a certain vicious type of ethnic cleansing, is proving to be an incurable propensity of the central German state.

[Nassim Nicholas Taleb @nntaleb]

The problem of lobbying & propaganda is that, in spite of costing a loooot of energy, they produce no cumulative gains. Once you stop the effort, everything reverts to the normal state and you end up as if you did nothing.
History is not linear; it zigzags but has a direction.

[Smiljana Skiba @MasaSkiba]

In general, when people selfstyle themselves "entrepreneur", "polymath", or something of the sort, they're not.
Real entrepreneurs calls themselves "restaurant chain CEO", "Ski equipment manufacturer", "kidney imaging importer". - NNT

[Codie Sanchez @Codie_Sanchez]

My dad always said he "sold marble and granite."

He never used the words Entrepreneur, Founder, or Owner.

There's a lesson there.