remorses

No support for multiple agents at the same time is a non starter

Using AI agents productively requires setting up a repository for collaboration, it means writing docs and making the build process easy and fast.

As any other tool AI is slow to adopt but has huge gains later on

Framer is one of the website builders with the best export capabilities, this is because it’s built on top of React. You can use unframer [0] to export components as React and ship anywhere

[0]: https://github.com/remorses/unframer

If you set temperature to zero the output will be always the same, not a distribution. If instead you increase temperature the LLM will sometimes choose other tokens than the one with the highest score, but it won’t be that much different

Insurance period minus one standard deviation

If only there was a way to make radio signal clear even with background noise. Maybe one day someone will invent frequency modulation instead of this stupid AM thing

Goblin is a service that builds a go binary for your platform on the fly and downloads it in PATH. This is a much faster way than setting up Github Actions to build an executable for every possible platform on every release. You can also use go install if you know what you are doing.

You can also install it with go, I updated the readme

go install github.com/remorses/docker-phobia

/skæm/ a dishonest plan for making money or getting an advantage, especially one that involves tricking people

Is the code open source?

Mojo is faster because it compiles every function body in parallel, basically every function calls LLVM separately. This means you have to do everything else yourself, which is not easy.

I am pretty sure the existing release process already takes much more than 2 hours.

It’s another one of my projects, holocron wasn’t ready when I made the blog

A license that doesn’t allow reselling the service is good enough

For example I saw this with Tailwind UI

== does not exist

It is specifically an alternative to the WorkOS Admin Portal component, a hosted website you can send to your customers to make setting up SSO easier.

The full source code is on GitHub: https://github.com/remorses/akarso

A spreadsheet should be more than enough

I cannot even push code

Using the CPU with quantized weights on GPT models makes sense, an example is llama.cpp, that’s because these models are constrained by memory bandwidth and not compute (low arithmetic density)

https://github.com/ggerganov/llama.cpp

It would be cool too see a port of llama.cpp and ggml to Zig

https://github.com/ggerganov/llama.cpp

https://github.com/ggerganov/ggml