[BUG] Got a connection error when I send a request
Expected Behavior
Returns a valid response and not a timeout error
Actual Behavior
Ends up by returning a timeout error. I am unsure if it's a bug or something I am doing wrong, but the example is simple, so maybe I am missing some parameters. The example is simple, so maybe I am missing some parameters.
Running on Docker Engine 20.10.14 on Mac OS.
Steps To Reproduce
#syntax=ghcr.io/sozercan/aikit:latest
apiVersion: v1alpha1
models:
- name: uncased-sentiment
source: https://github.com/lordofthejars/bert-base-multilingual-uncased-sentiment_onnx/releases/download/1.0.0/model.onnx
Then I build:
docker buildx create --use --name aikit-builder
docker buildx build . -t my-model -f aikitfile.yaml --load
Then I start it:
docker run -ti --rm -p 8080:8080 my-model
Finally I curl the model:
curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{
"model": "uncased-sentiment",
"messages": [{"role": "user", "content": "It is so good"}]
}'
And in the container console I got the following error:
rpc error: code = Unavailable desc = connection error: desc = "transport: Error while dialing: dial tcp 127.0.0.1:39863: connect: connection refused"
rpc error: code = Unavailable desc = connection error: desc = "transport: Error while dialing: dial tcp 127.0.0.1:39839: connect: connection refused"
rpc error: code = Unavailable desc = connection error: desc = "transport: Error while dialing: dial tcp 127.0.0.1:33679: connect: connection refused"
```
Any idea why this is happening? The example is a simple, not complicated lifecycle.
### Are you willing to submit PRs to contribute to this bug fix?
- [ ] Yes, I am willing to implement it.
@lordofthejars I am not sure if onnx format is supported by LocalAI yet. You can check the compatibility here: https://localai.io/model-compatibility/
Thank you very much. One question, do you know if any of these files https://huggingface.co/nlptown/bert-base-multilingual-uncased-sentiment/tree/main are supported by LocalAI. I end up loosing a bit between so many formats. Thanks.