> YouTubers upload about 720,000 hours of fresh video content per day. Over 500 hours of video were uploaded to YouTube per minute in 2020, which equals 30,000 new video uploads per hour. Between 2014 and 2020, the number of video hours uploaded grew by about 40%.
But what are you mostly "teaching" the LLM then? Mundane everyday stuff? I guess that would make them better at "being average human" but is that what we want? It already seems that prompting the LLM to be above-average ("pretend to be an expert") improves performance.
This whole conversation about training set size is bizarre. No one ever asks what’s in the training set. Why would a trillion tokens of mundane gossip improve a LLMs ability to do anything valuable at all?
If a scrape of the general internet, scientific papers and books isn’t enough, a trillion trillion trillion text messages to mom aren’t going to change matters.
I have the latter working on a M1 Macbook Air with very good results for what it is. Curious if bloomz.cpp is significantly better or just about the same.
I know from personal experience of making and selling an App Store app [1] that it can be difficult to show accurate pricing on your products website.
There can be 175+ different territories that you may sell an app in, each with a different local price point. You also can't expect to simply show a converted $US dollar amount either as you may want to do price discrimination (so that the app is cheaper for regions that cannot afford the exact $US equivalent).
Overall options seem to be:
1. Don't show pricing and leave it for the App Store to display - the most common approach
2. Only show $US pricing on the website and let non-US customers manually translate to their local pricing - not a great experience for non-US users
3. Manually or automatically sync pricing on the website display with the actual current App Store pricing. This is quite a labour or integration intensive task and I've only seen this been done by large vendors
Sounds like an excuse to just not display the price.
> 2. Only show $US pricing on the website and let non-US customers manually translate to their local pricing - not a great experience for non-US users
It's a worse experience for someone to convert USD to their currency (instant mentally or 2 seconds on Google) compared to launching the app store, installing the app, getting to the purchase menu, and seeing the price?
You don’t need to install the app, the prices are displayed on the product page in the App Store, towards the bottom. It’s listed automatically for all apps.
I believe that doesn't actually work in CSGO (possibly it works in local servers against bots, but not in multiplayer where people actually want to play).
Hence people making workarounds such as changing movement keys to both move and to execute the command that hides any already created blood (and bullets):
There are sites where you can purchase and download different world regions CD keys. I'd first remove CS:GO from your library completely (steam offers a way to do this) then activate the German version of the game. There is built in censoring for death/blood/corpse ragdolls afaik.
Let's take a step back. You have a model like linear regression. For example, y=bx where y are your outputs and x are your inputs. Based on some data, you learn that b=1. Therefore, you share the weights of the model as a file like {b=1} and also share with them the model y=bx (usually shared via code) so they can run it in production.
Your inputs are lists of numbers. Your outputs are lists of numbers. There exists some possible list of numbers such that, if you multiply your inputs by that list you'll get (approximately) the outputs.
In this conception that possible set of numbers are the weights. "Training" is when you run inputs, compare to known outputs, and then update the weights so they produce outputs closet to what you want.
Large Language Models, it may be hard to see how they fit this paradigm - basically convert a sequence to a list of numbers ('aardvark' is 1, 'apple' is 2 etc) and then the desired output is the next word in the sequence (represented as a number). Surprisingly, if you get good at predicting next word in sequence you also get the ChatGPT et al behavior.
This is what happens when running inference on a neural network:
Input (list of numbers) -> (Bunch of math operations) with (other numbers) -> Output (also a list of numbers)
This applies whether you are talking about image classification, image generation, text generation etc.
The model defines what the "(Bunch of math operations)" part is. As in, do these multiplications, then add, then a tanh operation etc.
The weights define what the "(other numbers)" are. Training is the process of figuring out these weights using various methods - some of which involve example inputs/outputs (supervised learning), others don't require examples (unsupervised or self-supervised learning).
> YouTubers upload about 720,000 hours of fresh video content per day. Over 500 hours of video were uploaded to YouTube per minute in 2020, which equals 30,000 new video uploads per hour. Between 2014 and 2020, the number of video hours uploaded grew by about 40%.