Viewing profile — eiz

https://arxiv.org/pdf/2302.13971.pdf table 15. 1770394 A100-80GB hours to train the entire model suite at the going rate for cloud 8xA100-80GBs (~$12/hr if you could actually get c…

> Where is the connection between computational details and the model's high-level behavior? Do we even know? This is an active area of study ("mechanistic interpretability") and i…

> 4. Describing positional embeddings as multiplicative. They are generally (and very counterintuitively to me, but nevertheless) additive with token embeddings. Worth noting that …

Fair enough. In the GH100 architecture doc https://resources.nvidia.com/en-us-tensor-core/gtc22-whitepa... (page 18) they do mention retaining 2 graphics-capable TPCs but it's clea…

> They do not have any graphics rendering circuitry. What? Not having a display output is not the same as not having graphics rendering circuitry. Here's vulkaninfo from an A100 bo…

Here's a couple more quotes from Yann LeCun, their Chief AI Scientist: https://twitter.com/ylecun/status/1643945742850031616 https://twitter.com/ylecun/status/1644503237699969026 p…

I tried this: > Solve the following word problem step by step: > "John is in a train moving west at ten miles per hour. Jill is in a train moving north at twenty miles per hour. Ho…

I've done this a few times too. > There’s no way to pre-register the pet. At least on Delta, you can. You need to call the airline and arrange it. > Oh, and to go through security,…

> This means no closures, which K implementers consider a feature (I don't). having not touched K in about 15 years, when did this change? in k3: K 3.2 2004-09-23 Copyright (C) 199…

> If you think Chat apps are winning, let me know when you can buy an item online without an account linked to an Email. I'd love to see an example. https://www.amazon.com/gp/help/…

We can do full (permanent) remote.

> It's an HDMI cable! The video signal has CRC in it and is packetized, it's either going to make it or it isn't. I don't disagree with your main point, but this actually isn't qui…

Last time I checked the "free" provisioning profiles it generates are valid for something like 7 days. It's not really convenient for anything but experimenting with iOS developmen…

The calculator you linked to assumes a maximum TCP window size of 64KB for everything but "replication". TCP window scaling has been on by default in every major OS for 10 years or…

Relevant: https://www.youtube.com/watch?v=Q0-51IkWpFE

house_arrest used to allow direct access to the documents and container of any app. It's used by the Xcode "installed applications" list, to allow you to download and upload contai…