Model Downloaded via Quick Usage vs. Docker-Compose Implementation

[hangartner8352]

I am new to this topic and have downloaded and implemented the code. The only adjustment I made to the docker-compose file was setting the model to large and the engine to faster_whisper. The model that was downloaded in this case was large-v3.pt. However, on a second machine, I used only the quick usage command:
docker run -d -p 9000:9000 -e ASR_MODEL=large -e ASR_ENGINE=faster_whisper onerahmet/openai-whisper-asr-webservice:latest
I noticed that it used a different model (models--Systran--faster-whisper-large-v3). Can someone explain why there is a difference?