Ettore Di Giacinto comments

Results 630 comments of


                                            Ettore Di Giacinto

[immucore] Drop mount libraries and use the mount syscall directly

Is this still relevant?

fix(reranker): llama-cpp sort score desc, crop top_n

> **Description** > > This PR fixes #7189 > > **Notes for Reviewers** > > After applying the fix I see the right reranker output: > > ``` > curl...

SSH Hardening

Thanks for filing the issue, this is something indeed we should support, either ootb or explicitly opting-out from the cloud config.

Must we need docker file to run LocalAI?

did you check the binary releases? https://github.com/mudler/LocalAI/releases/tag/v2.18.1

Deletion of configuration file results in deletion of model file even if other configuration file uses model file

Can confirm this, fix is on its way in #7235

fix(reranker): support omitting top_n

> #7025 introduced handling result cropping by `top_n`. However, I believe the most users just omit `top_n`, just because don't bother to count `len(documents)`. So, I'm afraid releasing #7025 causes...

Only use 4 CPU threads in P2P worker cluster

Hey @titogrima - LocalAI doesn't set any thread when running in `p2p` mode. This sounds more like a bug in `llama.cpp` as we just run the vanilla rpc service from...

Only use 4 CPU threads in P2P worker cluster

Also might be worth noting that you can pass any command options of llama.cpp from LocalAI with `--llama-cpp-args` or `LOCALAI_EXTRA_LLAMA_CPP_ARGS`, from the `--help` output: ``` ./local-ai worker p2p-llama-cpp-rpc --help Usage:...

chore: :arrow_up: Update ggerganov/whisper.cpp to `99b011a9f5e63f71201bfa583250506453a7b995`

mmh seems libwhisper.a was removed from Make targets, but bindings still refer to it :https://github.com/ggerganov/whisper.cpp/blob/37c88027e1cb7c4d8f9c7bc874f7da6d3e179984/bindings/go/Makefile#L33

fix(reranker): reproduce ignoring top_n

Hey, thanks for looking at this! > presumably it might be fixed for llama_cpp rerank as well, if @mudler consider it necessary. I think for llama.cpp we should be covered...