- You can access the model in Google Colab here: https://colab.research.google.com/drive/1fjKvM7X2oZc84I6dy-8_U_a3cRJNl5R9?usp=sharing
- You can access our website front end repo or our extension
- If you are interested in training the model. You can access it in the training folder
- You can build the container and run it locally assume you have GPU. This is for the backend only.
- You have to download dolly-v2-3b model locally and put it in app/models folder
- You can also change
PRETRAINED_MODEL_NAME_OR_PATH
toPRETRAINED_MODEL_NAME_OR_PATH = "databricks/dolly-v2-3b"
in app/main.py
- You can also change
- You can test it by editing the inferenceTest.json and then do curl command using that json file to 0.0.0.0:8080/predict
- You have to download dolly-v2-3b model locally and put it in app/models folder
- If you want to deploy on the cloud,
- upload the dolly-v2-3b model(https://huggingface.co/databricks/dolly-v2-3b) to Google Cloud Storage
- fix the first step in build.yaml to be your GCS bucket
- Then follow this document(https://cloud.google.com/vertex-ai/docs/general/deployment) to deploy to Vertex AI