Give it few shot examples and the quality should improve a lot. If you want a bot able to do technical Q&A give it two questions as well as the answers in its prompt. The model will then infer it's role as an AI that answer questions truthfully, and will try it's best to continue along those lines

I gave it the prompt thats the example from here: https://huggingface.co/blog/few-shot-learning-gpt-neo-and-in...

Tweet: "I hate it when my phone battery dies." Sentiment: Negative ### Tweet: "My day has been " Sentiment: Positive ### Tweet: "This is the link to the article" Sentiment: Neutral ### Tweet: "This new music video was incredibile" Sentiment:

and it's output was:

Tweet: "I hate it when my phone battery dies." Sentiment: Negative ### Tweet: "My day has been 90% balls." Seentiment: Negative ### Twitter: "i don't have a favourite type of music i just hate country" Seentiment: Negative ### Twitter: "c'mon first sunshine of the year...why the rain already?!" Seentiment: Negative ### Twitter: "Dear twitter, are you getting tired of me yet? Seentiment: Negative ### Twitter: "I don't care about the race, I just care about the president being a psych

Doesn't feel like that did the trick, but at least you got me to setup the shell command in a way where I pipe long-form prompts in via a separate file.

I think the model only has pre-training on large text but not multi-task fine-tuning and RLHF. So it can be improved a lot.

Hopefully a RLHF fine tuned model will be released soon because that will make the technology much more useful and less disappointing to people trying to run an alternative of ChatGPT on their local machine.

It invented "Seentiment"?

Just use iteration where you continuously update the prompt to include previous messages

Would probably pile up to an inhuman amount of data storage. Imagine having to pay for storing the equivalent of 1000 tokens of text within that budget of only 0.0002 dollars

That's one zero too many. Storage cost of 1000 tokens (6000 bytes) on a single HDD is $0.000000096 assuming $16/TB

The only way I could get it to say something was completely written by a human, was by adding spelling mistakes

Not really sure if this is supposed to be the joke, but your link just redirects me to google

So many people on HN just debate based on the title of an article that the actual writing of the content is a superfluous waste of time

Its a political choice in so far that we have artificially inflated home prices by not allowing enough homes to be build

The federal government spent 6.8 trillion about 30% of GDP in 2021. If we confiscated all wealth of every single billionaire in the US without any loss what so ever, we would have 4 trillion dollars, or enough to fund the budget for about seven months. In this case we are ignoring that you would probably only destroy that vast amount of wealth without having anything to show for it, because most of it is tied to things like stock prices, which you would probably have a hard time finding a buyer for if you just confiscated everyones stocks...

Moreover, could only do that $4 T confuscation once while the $6.8 T expenditures would continue on year by year.

Moreover, if just dumped the $4 T into ordinary consumer expenditures, i.e., houses, cars, Walmart, Amazon, etc., then would get some inflation which would mean that the $4 T would do less good than a simple view would have anticipated.

I will add that it's only about luck if your entering a certain kind of industry, where your succes is dependent on popularity.

If for example you start lumber company selling timber wood, the only thing you need to have an edge is to undercut the market with 5% on price. Of cause you won't get rich quick or maybe even slow, but it's a tried and tested strategy. Before starting a business you should ask yourself is my business luck depend, or is my success guaranteed as long as I undercut the competition on price or quality.

As a former poker player I can say it's quite stupid to use that kind of app at the table, because it's a massive tell. Any time somebody looks at that app the other people at the table know you have mid range hand. Cause your not gonna be looking at it if you have unsuited 2, 5, and you not gonna be looking at it when you have Pocket jacks. Reminds me of how one of the regulars I played with was one of the smartest people I knew, he always calculated the exact odds in his head while sitting at the table, but he was still a subpar player, because he had the most obvious tell, if he was staring kind of up and to the right it was because he was calculating his odds of hitting a range, meaning I could induce when I should bet to get him to leave the pool.

The fact that you think this is an obvious tell just shows that it could be used strategically against you. Some players will use any means available to level each other. Don’t be so sure :-)

Wouldn't an easy defense be to always look at the watch just to set a pattern?

My guess is we have to wait a good 3 months time before someone makes a opensource version

:)