This is a very interesting product! Congrats on the launch!
I was wondering if this would also be supported for no-SQL databases like Mongo?
Also a little sandbox to play with it would be nice. For example, you could have a small table of weather data from a random city, and people could then query something like "How many days in 2022 was the daily high for Los Angeles above 100 degrees?" Of course, this might be a lot of work as you would have to have it run on the backend and then return the results to the frontend.
Yes definitely, that is something that I think would be interesting for us to add. Since we are building everything fully on prem, we didn't have a web based product to confuse people, but I think your point stands that have a tangible demo to play with will help users trust the product to eventually run it on prem.
Speaking of AI generated pages, I wonder how OpenAI filter these low quality web pages out of their training set as they continue to training.
Also, I wonder how they decide what code is worth training on. Because a lot of code is written in poor style/has technical debt, it might be the case that these LLMs in the long run lead to an increase in the technical debt in our society. Plus, eventually, and this might already be happening, the LLM are going to end up training on their own outputs, so that could lead to self immolation by the model. I am not certain RLHF completely resolves this issue.
> I wonder how OpenAI filter these low quality web pages out of their training set as they continue to training.
This. The value proposition is very clearly tied to the quality of the training data, and if there's secret sauce for automatically determining information quality that's obviously huge. Google was built in part on such insights. I suspect they do have something. I'd be utterly astonished if quality sorting were an emergent property of LLMs (especially given it's iffy in humans).
The problem, of course, is that if they do have a way of privileging data for training, that information is going to be the center of the usual arms race for attention and thinking. It can't be truly public or it's dead.
yea i'm kind of shocked none of these models implement any kind of fingerprinting, something encoded in zero width spaces or other invisible unicode. It would be trivial to delete it but for the vast majority of cases, it would allow content to be flagged as model output-do not ingest
My dad moved to the US in 99 (he grew up in a town of about 50k people at the time). He said he was shocked people would stop for him when crossing the street, and he was amazed he could by a 1/2 gallon of ice cream for a couple bucks. Previously, he might have had ice cream once a year. He was also impressed and made happy by the abundance of meat and milk.
Funnily he was quite disappointed at the fruit situation coming from a tropical country, but that was more about freshness than quantity.
There's also the issue that people from tropical countries are used to a variety of fruit we simply don't have. There are various things that are either rare in the US or completely absent because they can't survive being shipped. Even subtropical--my wife loves dragon fruit but is unwilling to pay through the nose for the very inferior stuff in the market here.
While I do not have a taste for tropical fruit there are the bananas we encountered in Africa that ruined me. Since encountering those I have pretty much zero interest in US bananas, what we had there can't be bought in the US because they don't ship well.
I think the idea of "fully diluted market cap" is essentially bs. It is answering the question what if the supply went up but the demand went up to the same degree as to keep the price stable, what would the market cap be then?
One of the problems is price point. They are charging $10/month. While increasing the price will certainly decrease the number of users, a much higher price point could select for "power users" that will churn less and pay substantially more. In the long run that could lead to greater revenue.
Of course this depends on how the supply demand curve looks like for their specific business.
This is also about job security. Technical people often have a record of what they have done. Think commits and PRs. Managers don't have this. Their job doesn't produce anything tangible, this is not to say that it is useless. There are good managers and bad managers, with the majority falling in the latter category.
Being seen in meeting rooms filled with people, talking to them, and pointing to things on a powerpoint presentation can give other more senior managers (who determine their position and compensation) the aura that they are hard at work and making things happen.
My previous manager arranged the furniture in his office to be sitting facing the door, computer at his side, to be able to look at us, and call us, more easily.
Guess who was strongly for the return to office.
This reminds me of a podcast Ben Orenstein was on where he talks about something similar. Made a big difference in how I build/think about side projects.
It makes the job of the person referring you so much easier. If we agree that getting a referral is the easiest way to get a job, which I think it is, then making the job of the person referring you is the most important thing.
If they can talk about you for a couple of minutes and then email their boss a link to a well written blog post, you derisk the act of referring you. You allow the boss to sell himself by providing the material, so your friend isn't on the hook as badly.
I was wondering if this would also be supported for no-SQL databases like Mongo?
Also a little sandbox to play with it would be nice. For example, you could have a small table of weather data from a random city, and people could then query something like "How many days in 2022 was the daily high for Los Angeles above 100 degrees?" Of course, this might be a lot of work as you would have to have it run on the backend and then return the results to the frontend.