Smart search / facial recog. error : embedding.join() is not a function
Hello all, hope you are well.
Firstly, I am a newbie in unix and self hosting overall. I recently installed immich-server on my Rasp 4, working really good with the interface and uploads etc.
Until then, I had deactivated immich machine learning as my poor RPI 4 wouldnt be able to handle it. Instead, I chose to install it on my big gaming setup rolling on RTX 3060 ti, i7-10700K Win 11.
Right now, Immich-ML is running on docker on windows with WSL, logs seem fine. It downloaded the models requiered as set on immich server on its own. As I try to load it with facial recognition tasks, I see that it receives a consequent amount of data and also returns some, which makes me believe that the immich ML side works properly.
However, as I trigger a smart search or a facial recognition from the interface (so from immich server), nothing happens back (search fails, facial recognition shows no change) even though packages were sent and the process seem to have happened immich ML side.
Firstly, I am a newbie in unix and self hosting overall. I recently installed immich-server on my Rasp 4, working really good with the interface and uploads etc.
Until then, I had deactivated immich machine learning as my poor RPI 4 wouldnt be able to handle it. Instead, I chose to install it on my big gaming setup rolling on RTX 3060 ti, i7-10700K Win 11.
Right now, Immich-ML is running on docker on windows with WSL, logs seem fine. It downloaded the models requiered as set on immich server on its own. As I try to load it with facial recognition tasks, I see that it receives a consequent amount of data and also returns some, which makes me believe that the immich ML side works properly.
However, as I trigger a smart search or a facial recognition from the interface (so from immich server), nothing happens back (search fails, facial recognition shows no change) even though packages were sent and the process seem to have happened immich ML side.
logs, docker compose and env down below