bllchmbrs

I've thought about this problem set for years, I've written many docs and technical books.

Version control is the best for documentation.

But maintaining it is hard - lots of great comments here.

For anyone interested, I'm working on https://hyperlint.com/ (disclaimer: bootstrapped founder). To help automate the toil around documentation.

As more and more products integrate AI, this kind of testing is going to get more and more critical.

This was my project that I recently sold. If you want to be listed I can put you in contact with the folks that bought it. LMK!

The lack of "contact us" was pure laziness at the time.

For those that wants prices checkout: https://gpumonger.com/

This seems much more in depth, and a true service, but for those that just want to compare prices check out gpumonger.

Confusing pricing for Kimai. Start for free for hosted but I can't find the pricing anywhere on the website but I found a blog post about how it's no longer free. Do you pay for it and try to host it?

Check out other prices on https://gpumonger.com/

Disclosure: I collected the data and built the site, but it has a ton of comparison data for GPU clouds.

I'll fix this, thanks for the feedback.

Thanks for the feedback!

that's why I built: https://gpumonger.com/

You can see all alternative GPU providers: https://gpumonger.com/cloud-gpu-providers as well as the main ones.

Oracle has great GPU prices: https://gpumonger.com/cloud-gpu-provider/oracle-cloud

For those that want to see where you can try an L40 in the cloud checkout: https://gpumonger.com/gpu/l40#providers

There are 2 problems with the analysis in the article:

1. They choose startups from 2014-2015. Valuations were (a) more reasonable then and (b) we've just been through one of the biggest bull markets of tech stocks in the past decades. 2. That taking the "Mean Valuation Growth" over that time period is meaningless if (a) there hasn't been a liquidity event and (b) if the mean is weighted by outliers (which it always is).

Portfolio / VC strategy is a bet on power laws. We, as employees, don't have that opportunity. That means we've got to be deliberate about who we join to try to hit a power law outcome.

Am I crazy here?

DataFrames are just SQL. There will be no performance difference.

RDDs will be worse, so it shouldn't matter. No vectorization, no column processing, lots of serialization and de-serialization. They're basically always slower than DataFrames barring some strange use case.

Seriously, why isn't anyone else talking about this. They turned this around SO FAST.