Oooo nice, not really yet... Claude stores your session history in ~/.claude/projects/-Path-to-git-repo on your local machine. I should definitely add a "teleport" command like the one claude itself has which let's you move sessions between machines. It's a little tricky with git state & un commited changes but definitely possible. I'll consider this an upvote for that feature.

(1) This one is on the top of my list. I have a branch that sets up proper iOS notifications. Coming soon. (2) Yeah, I've considered making an actual desktop version. A hacky one exists already. You can run `catnip run` from your desktop and it will run catnip in a Docker container on your machine. There's a web interface, and it even works with native MacOS notifications.

Re: #2, I think I was looking for something different and likely simpler than what you were thinking... The session would still be running in a codespace, I just want to be able to start on my phone, then move to my Mac, which has a nicer keyboard, and still have the codespace automatically shut down when things go idle (depending on how you have that set up, it may be that I need a dedicated app for that, or it may be I just need to VA Code remote into the codespace).

Ahh, understood. Today you can install the iPad version to your mac, not the best experience but would get you what you want. I'll do some more thinking about a more Mac OS native experience.

Tailwind is especially well suited for LLM generation because all of the styles come inline with the HTML. Could definitely just ask it to write styles but this approach also improves token efficiency.

https://nsjail.dev is a common tool for sandboxing

A100’s are Nvidia GPU’s. You can rent them from providers like AWS or LamdaLabs. The readme has instructions for downloading the original GPT2 weights from OpenAI. You can also train a very simple version on a smaller dataset from your laptop as described in the README.

If you just want to play with a similar but much better model goto https://chat.openai.com

Pretty cool this post is already the first result in Google’s index. How does that work?

This would be awesome for terraform! It could potentially derive the details of SSL, k8s, Load Balancers, and databases for each cloud from comments

Hey Lenny! Congrats on the launch, so excited to see your product come together.

Thank you Chris for your advice and support! The partnership you and Luke have is an inspiration for Ross and I.

My bad, this was using non-Accelerated TensorFlow on a 2.3GHz 8-Core i9.

GCP charges ~$0.60/hour for the GPU/CPU/MEM equivalent in colab. If bandwidth is costing 10k per day, how much is the free colab compute costing Google?

Probably less. I doubt the average use time is more than half an hour. Pretty cheap advertising for the service, really.