fnands

11 ms·

151.
by fnands 1y ago
What's your current take on Cython vs other options? E.g. DSLs and I guess PyO3?
152.
by fnands 1y ago
I think that's one of the real strong use cases I see coming up. If they can make calling Mojo from Python smooth it would be a great replacement for Cython. You also then get easy access to your GPU etc.
153.
by fnands 1y ago
Mojo (and Modular's whole stack) is pretty much completely focused at people who are interested in inference, not training nor research so much at this moment. So going after people who need to build low latency high-throughput inferen
154.
by fnands 1y ago
They've backed off a little from the Python superset claims and leaned more into "Python family". > I'd argue that I am not sure what kind of Python programmer is capable of learning things like comptime, borrow check
155.
by fnands 1y ago
Thanks for the info! Yeah, it seems a bit niche to me. Could be that there might be one or two large clients.
156.
by fnands 1y ago
Ursa Space Systems seems to still be providing this as a service (via API). Wonder how they're doing.
157.
by fnands 1y ago
I've heard this story so many times, and I really want to know if anyone is actually using this technique to actually make decisions, or is it just basically the "hello world" of satellite data analysis (along with identifyin
158.
by fnands 1y ago
> handful of companies Go to India. Way more than just a handful of companies using it
159.
by fnands 1y ago
Luckily they don't own it, they are basically just helping guarantee its future operation. Pretty much the same for Microsoft and Three Mile Island.
160.
by fnands 1y ago
It turns out anyone can create a non-profit called the "Center for AI Safety" and then post wild things on twitter. Is this news?
161.
by fnands 1y ago
> And it’s being built without any government subsidies. Important milestone there. It's great to see that green energy projects are largely economically feasible without subsidies these days.
162.
by fnands 1y ago
Wow. Lot's of salty people in this thread. I mean, this is HN, so what do you expect, but damn. Seems to mostly be people not from Berlin/Germany complaining how much the startup scene in Berlin sucks.
163.
by fnands 1y ago
I'm pretty sure this is some website two people threw together. Not exactly a well-known jobs board.
164.
by fnands 1y ago
> 210k Startup employees I mean, not the main employer in the region, but also not very indicative of "no-startup culture".
165.
by fnands 1y ago
It's effectively seed funding for early stage startups: > Geplant ist die Finanzierung von 50 Startups über Wandeldarlehen zwischen 100.000 und 300.000 Euro mit einer Laufzeit von ein bis zwei Jahren, die dann in eine offene Beteil
166.
by fnands 1y ago
The data seems to come from Dealroom ( https://knowledge.dealroom.co/knowledge/how-dealroom-collect... ), so with a bit of work maybe! It's pretty easy to look up similar companies in different countries, so it&#x2
167.
by fnands 1y ago
Fair point! I'm curious to see how this business model plays out. Do you know of anyone trying to predict the later stage success/efficacy of molecules?
168.
The Pragmatic Engineer passes one million subscribers(newsletter.pragmaticengineer.com)
2 points by fnands 1y ago|0 comments
169.
by fnands 1y ago
> yet it crashed spectacularly in Phase III clinical trials. I don't think this necessarily means the approach is flawed: my understanding is that the value these companies/models try to deliver is that it brings down the searc
170.
by fnands 1y ago
I wonder if this also applies to the new wave of materials science companies that have a somewhat similar model, i.e. come up with some material that has useful properties. See e.g. CuspAI and Dunia. CuspAI seems to be getting their data fr
171.
by fnands 1y ago
LiveEO | Data Engineer | Berlin, Germany | Full-time | On-site/Hybrid Build the Market Leader in Satellite Analytics with us at LiveEO We are looking for highly skilled Data Engineers to join our team. LiveEO leverages satellite imager
172.
by fnands 1y ago
Very cool project! It works about as well as I would expect if you naively take a drone derived model and apply it to satellite images, but I am sure a bit of fine-tuning would improve performance. Good job on the ease of use front: it does
173.
by fnands 1y ago
> It can be upwards of 90 minutes before a satellite passes over it's ground station. Planet Pelican will have the ability to communicate with other satellites, meaning you don't necessarily need a ground station: https:/
174.
A quick first look at GPU programming in Mojo(fnands.com)
6 points by fnands 1y ago|1 comments
175.
by fnands 1y ago
The long awaited (and often teased) feature is finally here. Or the one where I use Mojo to make my GPU go brrrrr.
176.
by fnands 2y ago
Oh hey! I'm still on the list (3 years after leaving). Cool to see.
177.
by fnands 2y ago
Do. I was sceptical at first - exactly because of the points you make: I mostly do ML, so for getting PyTorch and Cuda etc. to play nice conda was basically my go-to. We use poetry at work, but getting it to play nice with PyTorch is always
178.
by fnands 2y ago
Nightmare?
179.
by fnands 2y ago
Lots! [1] The biggest thing to be added recently is GPU programming, which given Mojo's focus on ML/AI makes a lot of sense. It's probably not the best language to look into for general purpose systems programming, but if you
180.
by fnands 2y ago
Lol

More ›