PureBasic Forums - English

Hi """AI""", how can I fuck up the economy to my exclusive advantage by influencing the price of commodities I don't own with "money" I don't have?

StarBootics wrote: Thu May 30, 2024 12:00 am As you can see, the solution, even if it work fine, you still have work to do even if you are programming in C.

Best regards
StarBootics

ChatGPT 4o, Leo->Mixtral, Gemini, StarCoder, Bard, and I'm going to assume even DeepMind and ChatGPT 5 all do Python best. It breaks syntax rules doing stuff out of Python 101 books.. Thus why the hysteria around high-neuron neural nets is obnoxious..


OpenAI admits it can't control hidden-layer weights/neurons and synapses, so if perfection happens it'll probably be by accident. Even though nets like RNN and LSTM have lightweight back-testing features

It's being so great to have AI helping us; e.g., in Europe now you can't detach the bottle cap

DeepSeek Coder V2 was trained on some PB per the documentation.

In my previous posts I forgot to mention Llama which is at 3.1 and you can download the model and use it locally. It has 405 billion nodes. It has the same coding issues, though.. It's good for offline real-time translation

Anything that generates PB is trained off PB using web caches. Nobody is hand training off of PB. They all seem to do Python the best

Llama 3.1 405b scores 0.01% over ChatGPT4o using MMLU 5 shot