localai 2.18.1 (new formula)

[mauromorales]

Adds formula for LocalAI https://localai.io/

• Have you followed the guidelines for contributing?
• Have you ensured that your commits follow the commit style guide?
• Have you checked that there aren't other open pull requests for the same formula update/change?
• Have you built your formula locally with HOMEBREW_NO_INSTALL_FROM_API=1 brew install --build-from-source <formula>, where <formula> is the name of the formula you're submitting?
• Is your test running fine brew test <formula>, where <formula> is the name of the formula you're submitting?
• Does your build pass brew audit --strict <formula> (after doing HOMEBREW_NO_INSTALL_FROM_API=1 brew install --build-from-source <formula>)? If this is a new formula, does it pass brew audit --new <formula>?

---

[chenrui333]

still some test failure

  ==> curl -s -i 127.0.0.1:49482
  
  Killing child processes...
  Error: localai: failed
  Error: localai: failed
  An exception occurred within a child process:
    Minitest::Assertion: Expected: 0
    Actual: 7

[prabirshrestha]

Would be great if the package was update to use 2.16.0 as it supports new features such as function calling and also a single binary release so doesn't require compilation. Though I added to explicitly run brew install abseil grpc for the text generation models to work. Ideally those 2 packages should probably be statically linked in LocalAI.

[SMillerDev]

Ideally those 2 packages should probably be statically linked in LocalAI.

Nothing should be statically linked in Homebrew formulae.

[mauromorales]

@chenrui333 I don't really have such systems (macos 14 x86_64) to debug, I find it interesting that it works on macos 12 with the same arch, do you think it's a matter of increasing the delay a bit more? Or worst case scenario, can I say it's just arm64 available, since it's the only OS I myself can test?

[mauromorales]

@prabirshrestha I'd rather have this merged first than introducing even more changes, as soon as it is merged we will get a version bump and we can see if there's much needed to change or if it works as is.

But maybe version 2.16 passes the tests and then it's worth doing already 🤷

[mauromorales]

@carlocab do you know why I encounter this issue on the runners but not on local?

CMake Error at /opt/homebrew/Cellar/cmake/3.30.0/share/cmake/Modules/CMakeDetermineCompilerABI.cmake:129 (message):
    The CXX compiler targets architectures:
  
      "arm64"
  
    but CMAKE_OSX_ARCHITECTURES is
  
      "arm64;x86_64"

[mauromorales]

Couldn't figure out where the issue comes from, but thankfully it was possible to remove one of the backends in LocalAI and this made it work.

[mauromorales]

@chenrui333 tests are green now

[mauromorales]

@SMillerDev would you have some time to review this?

[chenrui333]

do we have issue tracking this?

[mauromorales]

@chenrui333 issue in which project you mean?

[chenrui333]

I guess whatever project that we can followup on this :) I was thinking localai

[mauromorales]

It's not an issue of LocalAI itself, the gpt4all backend can only be built one arch at a time, so passing the multi-arch will always break. I don't think this should block this PR from being merged though. I'm already working for a cleaner way to exclude backends in LocalAI, which I can use once the next release is out, but it would be very helpful if we could have the formula already merged

[SMillerDev]

But why does it need to pass multiarch? Brew also only builds everything for a single arch at a time.

[mauromorales]

@SMillerDev good question, I'm not sure why the runner has CMAKE_OSX_ARCHITECTURES set to "arm64;x86_64" which is why this error pops up

CMake Error at /opt/homebrew/Cellar/cmake/3.30.0/share/cmake/Modules/CMakeDetermineCompilerABI.cmake:129 (message):
    The CXX compiler targets architectures:
  
      "arm64"
  
    but CMAKE_OSX_ARCHITECTURES is
  
      "arm64;x86_64"

sorry not a macOS expert here, so cannot really tell if this is expected in general or just on the runners

[mauromorales]

I tried in different ways to set CMAKE_OSX_ARCHITECTURES just the currently running arch, but that didn't work. I think that is related to how the shell out in ruby does the passing of the environment variables

[SMillerDev]

It's not supposed to be an environment variable, how about setting -DCMAKE_OSX_ARCHITECTURES=#{Hardware::CPU.arch} as a CMake arg?

[chenrui333]

@mauromorales can we try out the cmake arg route?

[mauromorales]

@carlocab @chenrui333 @SMillerDev tests are now passing, would it be possible to remove the "test failure" tag?

[AyushyaChitransh]

Is there anything pending on this formula? I would love to help on this to take this to completion.

[carlocab]

Waiting on response to #172734 (comment)

[github-actions]

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

[AyushyaChitransh]

Hello @carloca @chenrui333 @SMillerDev

The tests are passing for this PR. Awaiting for you inputs.

[SMillerDev]

See #172734 (review)

[prabirshrestha]

Seems like gpt4all backend is getting deprecated in LocalAI so probably that comment should be removed instead.

https://github.com/mudler/LocalAI/releases/tag/v2.20.1

[github-actions]

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

[chenrui333]

trying to build the latest 2.20.1 release, but seeing some of the following build errors:

backend.protobackend.protobackend.proto::11backend.proto:10:: 1Unrecognized syntax identifier "proto3".  This parser only recognizes "proto2".::1
:1010: Unrecognized syntax identifier "proto3".  This parser only recognizes "proto2".:
Unrecognized syntax identifier "proto3".  This parser only recognizes "proto2".:
10: Unrecognized syntax identifier "proto3".  This parser only recognizes "proto2".

[chenrui333]

Seems like gpt4all backend is getting deprecated in LocalAI so probably that comment should be removed instead.

https://github.com/mudler/LocalAI/releases/tag/v2.20.1

• chore: drop gpt4all.cpp mudler/LocalAI#3106