“What I cannot create, I do not understand.”
The quote from Richard Feynman raises a question for vision-language models:
Can VLMs fully understand the visual world if they cannot generate it?
VLMs are built for visual understanding, but they have structural limits. In this blogpost, @MRBarhdadi explores whether the strengths of generative models can transfer to VLMs and help close that gap.
"dumb it down for me a bit more, you can use up to double the words"
this is a prompt i find myself going back to a lot. Always asking the user to be brief, then pulling this bad boy out when the answer is too dense.
Am I the only one who thinks this Fable ban will be lifted by Tuesday?
I’m more concerned that it’s precedent setting for the government having its finger on the AI switch going forward.
if you hated how the open internet turned out, you have a chance now to fight against the next wave or corporate monopoly, shame on you if you are fooled twice
Tin foil hat time
Once anthropic can make money from their own software without selling the model but selling services created by their model they will stop being a model provider.
They are using you and selling to you as a stop gap before "AGI"