Helm chart for deploying CodeGen service.
CodeGen depends on LLM and tgi microservice, refer to llm-uservice and tgi for more config details.
To install the chart, run the following:
cd GenAIInfra/helm-charts/
./update_dependency.sh
helm dependency update codegen
export HFTOKEN="insert-your-huggingface-token-here"
export MODELDIR="/mnt/opea-models"
export MODELNAME="meta-llama/CodeLlama-7b-hf"
# To run on Xeon
helm install codegen codegen --set global.HUGGINGFACEHUB_API_TOKEN=${HFTOKEN} --set global.modelUseHostPath=${MODELDIR} --set tgi.LLM_MODEL_ID=${MODELNAME}
# To run on Gaudi
#helm install codegen codegen --set global.HUGGINGFACEHUB_API_TOKEN=${HFTOKEN} --set global.modelUseHostPath=${MODELDIR} --set tgi.LLM_MODEL_ID=${MODELNAME} -f codegen/gaudi-values.yaml
-
To use model
meta-llama/CodeLlama-7b-hf
, you should first goto the huggingface model card to apply for the model access first. You need to make sure your huggingface token has at least read access to that model. -
Make sure your
MODELDIR
exists on the node where your workload is schedueled so you can cache the downloaded model for next time use. Otherwise, setglobal.modelUseHostPath
to 'null' if you don't want to cache the model.
To verify the installation, run the command kubectl get pod
to make sure all pods are running.
Then run the command kubectl port-forward svc/codegen 7778:7778
to expose the service for access.
Open another terminal and run the following command to verify the service if working:
curl http://localhost:7778/v1/codegen \
-H "Content-Type: application/json" \
-d '{"messages": "Implement a high-level API for a TODO list application. The API takes as input an operation request and updates the TODO list in place. If the request is invalid, raise an exception."}'
Key | Type | Default | Description |
---|---|---|---|
image.repository | string | "opea/codegen" |
|
service.port | string | "7778" |
|
global.HUGGINGFACEHUB_API_TOKEN | string | "" |
Your own Hugging Face API token |
global.modelUseHostPath | string | "/mnt/opea-models" |
Cached models directory, tgi will not download if the model is cached here. The host path "modelUseHostPath" will be mounted to container as /data directory. Set this to empty/null will force it to download models every time. |
tgi.LLM_MODEL_ID | string | "meta-llama/CodeLlama-7b-hf" |
Models id from https://huggingface.co/, or predownloaded model directory |