Train your model

You can download a sample dataset here. Make sure to extract the contents of the file in the data directory. To do so, run

tar -zxf cats_and_dogs.tar.gz -C data/

Now you are ready to train your model. First set up the environment running

poetry install

Then, prepare the dataset by creating the 3 splits

poetry run python -m animal_classifier dataset-split data/train.csv data/val.csv data/test.csv --train-frac 0.6 --val-frac 0.2 --test-frac 0.2

Finally, train your model by running

poetry run python -m animal_classifier training data/train.csv data/val.csv --annotations-dir data/cats_and_dogs/annotations --model-dir models/cats_and_dogs

Evaluate your model

You can evaluate your model against a baseline by running

poetry run python -m animal_classifier evaluation base_model_latest.pth base_model.pth data/test.csv --annotations-dir data/cats_and_dogs/annotations --model-dir models/cats_and_dogs

Validate your model

You can also verify your model's performance is below a threshold by running

poetry run python -m animal_classifier validation base_model.pth data/test.csv --annotations-dir data/cats_and_dogs/annotations --model-dir models/cats_and_dogs --max-loss-validation 5

Serve your model

First you need to package your model. To do so, execute the following command from apps/animal_classifier:

poetry run torch-model-archiver --model-name animal --version 1.0 --model-file animal_classifier/models/model.py --serialized-file ./models/cats_and_dogs/base_model.pth --handler animal_classifier/api/torchserve/handler.py --export-path ./model_store/

Then, you can serve your in a Docker container. Go back to the root directory and run:

./scripts/serve_torchserve.sh

You can check the model is running by executing the following command:

curl http://localhost:8081/models/animal

Finally, you can make predictions using the following command:

• Single request

  curl http://localhost:8080/predictions/animal -T ./data/cats_and_dogs/frames/1.png

• Concurrent requests for batch processing:

  curl -X POST http://localhost:8080/predictions/animal -T ./data/cats_and_dogs/frames/1.png & curl -X POST http://localhost:8080/predictions/animal -T ./data/cats_and_dogs/frames/2.png

Serve your Web API

You can serve your in a Docker container. Go back to the root directory and run:

./scripts/serve_fastapi.sh

You can check the model is running by executing the following command:

curl http://localhost:8081/models/animal

Finally, you can make predictions using the following command:

curl -X POST http://localhost:8000/predict -H 'Content-Type: application/json' -d '{"base_64_str": "your_base64_string_here"}'

To update config "http://localhost:8000/api/setconfigtraffic?val=75"