Confirmed that mini uses ~30x more tokens than base gpt-4o using same image/same prompt: { completionTokens: 46, promptTokens: 14207, totalTokens: 14253 } vs. { completionTokens: 82, promptTokens: 465, totalTokens: 547 }.
Page 7 of their technical report [0] has a better apples to apples comparison. Why they choose to show apples to oranges on their landing page is odd to me.
We've been using Braintrust for evals at Zapier and it's been really great -- pumped to try out this proxy (which should be able to replace some custom code we've written internally for the same purpose!).
This was a ton of fun to build! We'll also be releasing a NLA enabled version of our Chrome extension [0] within the next couple of days which will be similar (but way more convenient than) the demo on the landing page above.
We're super bullish on LLMs for pulling "no-code" forward, helping more knowledge workers build automations. Already, folks are using our OpenAI [1] + ChatGPT [2] integrations to build very cool automations with summaries, categorization, copy writing, and more. We think there is a ton more to do here.
If anyone is interested in this problem space, shoot me an email [email protected]!
This isn’t exactly true. The Django ORM can be used with care in the async views found in FastAPI and Django (see sync_to_async and run_in_threadpool helpers).
Plans exist to make the Django Queryset async, so it’ll be exciting when that day comes!
I hope it continues to catch on! Major metros certainly have a lot to offer, but minor metros or rural America does too -- it's a little odd that all the online "tech" jobs land in the former. If you have a decent internet connection, you can do your work from anywhere.
Just peeking through the tracker history https://bugzilla.mozilla.org/show_bug.cgi?id=1397874 and you can see the effort it took to land this magnitude of a change -- pretty astounding. Big hat tip to Brian and the FF crew!
Zapier is 100% remote from YC S12. We're just over 7 years old, wrapped 2018 with 200 folks across 20+ countries -- about 75 of those are engineers! We're sustainably profitable and have a lot of cool stuff we want to do in 2019 to bring automation to the masses.
If you haven't heard of Zapier, we're sort of like Legos for the internet -- we help people build workflows around SaaS software or other online productivity tools. We're doing well, but we have a lot of work to do still. I think 2019 might be Zapier's most exciting year to date! :-)
We're also doing a lot of technical investment in 2019 around simplifying and scaling the product, evolving the architectural design, and of course a bit of tech debt reduction (which is mostly deleting code without breaking anything!).
We're also going to be hiring a Director/Manager-of-Managers layer as well as some Security Engineers in early 2019, but the job posts aren't up yet. If you are interested in those, maybe drop me a line [email protected]?
One thing that is striking to me in almost all these sorts of otherwise impressive demonstrations are the apparently bizarre "jitter" movements while waiting for a door to open or path to clear in the game. Clearly there is no fitness in quietly waiting.
It is darkly humorous to contrast Hollywood's or scifi's "killer AI robots" that methodically hunt you down to these real world demonstrations of emerging AI. Maybe the first "killer AI robots" would exhibit similarly bizarre behaviors while they methodically hunt down the unlikely hero. :-)
We certainly don't claim to have all the answers, but we've found it makes people pretty happy and we all seem pretty productive. If you want to give it a go, we're hiring! ;-) https://zapier.com/jobs/
The great YouTube channel "PBS Space Time" did a video on Gaia and how it is changing astronomy: https://www.youtube.com/watch?v=Jdy09y0A4t0. I highly recommend the channel. I really appreciate that they don't pull any punches, you'll often see math formulas on screen! Most astronomy education material (think classic Discovery channel) tends to be extremely watered down...