I ended doing a similar thing, but each tg bot can choose what repo/session to attend to.

https://github.com/botverse/tgcc

I found that cc is all you need indeed

On top of that is something they should have had from earlier times. My biggest pain point is to not to be able to continue from my phone. I just use a service to pipe telegram to any cc session in the dev machine. This is the number 1 reason why I got excited by openclaw in the first place but its overkill to have it just to control cc

Same here, I’m vibecoding a toy project where I never looked at the code from my phone, but I always seat for work. I’m using happy app and that’s good enough for now, I have the desktop in tailscale but I access it that way just for testing

I did something similar with copy until I found this which works across remote terminals too:

`alias clip="base64 | xargs -0 printf '\e]52;c;%s\007'"`

It just sends it to the client’s terminal clipboard.

`cat thing.txt | clip`

Having a decade of fresh air is also a good incentive regardless of how it ends

I don't know, it is a lot of effort for a decade fresh air. Then you will notice same policies implemented since they will take reference to how people solved it in the past.

I’ve been thinking about how to teach the size and proportions of the solar system to my kids, I’ve bought a couple of packs of blank RFID cards on which I intend to paint the planets over a starred background. And then walk with my kids the meters necessary to cover the distances before displaying them. What I don’t know is if there is a clever way to use the RFID tech, this website kinda offers an idea.

Or you could come to Sweden :)

https://en.m.wikipedia.org/wiki/Sweden_Solar_System

space engine is neat

I can’t seem to reconcile this with the fact that there has not been a significant improvement of the transformers since 8 years ago. None of the AGI components, in addition to those that emerged with hyper scale, have been achieved. Where is the superintelligence going to come from?

I have stopped following Matt Berman, his enthusiasm and love for OpenAI has unfortunately made him lose all semblance of critical thinking when it comes to this company, so his videos are too aligned with OpenAI marketing...

He's absolutely very enthusiastic about many things! I take that into account and mentally adjust when listening to him, but he's been great as [one] source of AI product news.

#378

I'll write a guide "no-code LLMs in CUDA".

This is most likely because ChatGPT returning the actual token corresponding to `<|endoftext|>` instead of returning the text representation of it

So it’s not crashing.

Try it for yourself, it enters in a loop, so it probably detects the abrupt termination and tries to start again, not sure how to call that behaviour, but crashing is not too incorrect in that case

Oh ok, I couldn't tell that from the screenshot.