Thanks! I added the check marks because when I was testing different quantizations, I often picked a model and only afterwards discovered that it couldn’t even load — it just didn’t fit into RAM or VRAM.
So the check mark simply indicates that the model can actually run under those constraints (fits in memory), not that it’s selected.
[…]the model’s internal representations become more aligned with human neural activity after fine-tuning. Taken together, our results demonstrate that it is possible to discover computational models that capture human behaviour across a wide range of domains.
Hi Eric, my name is Francesco. In the last three years we industrialized a novel method for producing biosimilar insulin (in Italy). We are an international team and I really would like to contact you. Can you write to the email you find on my profile?
Quite unknown, I discovered it more than 10 years ago. It is an easily and pleasantly readable book: I discovered my stupid actions, and I still use it to evaluate what I am going to do.
It isn’t a survival guide
or a “how to” manual, but it offers practical solutions to improve those human qualities that counteract stupidity.
Entropy, energy and life was the topic of a very interesting book I read in the last two days: “why information grows” by Cesar Hidalgo. If you are interested on this topic it’s a must read!
I try to take a different direction. Let's suppose some ai generated content is better than human created content (IMHO we are quite there). Let's go further: maybe in the future we will trust again only some "trusted sources" (newspapers? HN?) while everything else will be not taken into account because the quality will be low (like some comments saying the source is not in the industry...).
Thank you for your post and thanks for all the comments... it takes me years back. Maybe a dumb question: why we all miss those days? I don't think it is only related to being young. I am convinced it is something related to a "less noise/new tools" and probably I agree about "the more constraints more creativity". Is there anybody who can better articulate or have a different vision?
I agree with you, I have been involved in startup and patents, and the feeling was similar. I can add that patents can also be viewed as a "freedom to operate" certificate: lower chances that someone else claims you can't do what you are doing.
My very first (and maybe not so useful comment): I really love your approach. I was thinking something more automatic (I have some withings stuff) but you convinced me to start small and manual. Really interesting. Thank you for sharing.
So the check mark simply indicates that the model can actually run under those constraints (fits in memory), not that it’s selected.