Appendix : ALO-v3 CLI Command Guide
This section explains the CLI command list provided by ALO v3.
1. alo example
alo example command provides a simple example for quick run. This example uses Titanic data to predict the number of survivors, providing all the necessary files for AI Solution registration. Users can use this to easily register an AI Solution to AIC. Execute as follows:
alo example
2. alo run
The alo run command is a CLI that converts the user-written code into an ALO format. This CLI should be executed in the path where the user-written .py file and experimental_plan.yaml are located. The arguments that can be modified when running alo run are as follows:
| Args. Name | Type | Default | Possible Values | Description |
|---|---|---|---|---|
| name | str | - | - | Name of the solution |
| mode | str | all | ['train', 'inference', 'all'] | ALO mode: 'train', 'inference', 'all' |
| loop | - | - | - | Loop (no attribute specified) |
| computing | str | local | ['local', 'daemon'] | Training resource |
| config | str | - | - | Path of experimental_plan.yaml |
| system | str | - | - | Path of solution_metadata.yaml |
| log_level | str | ERROR | ['NOTSET', 'DEBUG', 'INFO', 'WARNING', 'ERROR'] | Log level |
alo run --log_level INFO
3. alo history
The alo history command provides a simple way to review the records of experiments conducted so far in the terminal window. This process provides various information such as the performance of the code written so far and the changed parameters. The arguments that can be modified when running alo history are as follows:
| Args. Name | Type | Default | Possible Values | Description |
|---|---|---|---|---|
| config | str | - | - | Path of experimental_plan.yaml |
| head | int | - | - | Output the last part of history |
| mode | str | ['train', 'inference'] | ['train', 'inference'] | Selection of mode |
| tail | int | - | - | Output the first part of history |
alo history --head 5 --mode train
4. alo register
The alo register command performs the process of registering the code written in ALO format, along with the experimental_plan.yaml and infrastructure information, in AIC. The CLI command takes the AIC ID and password as arguments. An example is written below:
Constraints:
- Uppercase English letters, spaces, special characters, and Korean are not supported, and up to 50 characters are allowed.
- The AI Solution name should consist only of lowercase English letters, dashes (-), and numbers.
alo register --id gildong.hong --password gildong.hong
5. alo update
The alo update command provides a convenient way to update an existing AI Solution with the same name in AIC. AIC ID and password are required when running the CLI command. An example is written below:
alo update --id gildong.hong --password gildong.hong
6. alo delete
The alo delete command allows users to easily delete a registered AI Solution. AIC ID and password are required when running the CLI command. An example is written below:
alo delete --id gildong.hong --password gildong.hong
7. alo docker
The alo docker command is used when users need to modify the Docker file. When this command is executed, a Docker file is created in the user's current path. When alo run is executed in the path containing the Docker file, it will run based on the Docker file in the current path.
alo docker
Add the --gpu argument if you want to use a GPU.
alo docker --gpu
The part to change the CUDA version in the generated Docker file is as follows:
## torch >= 2.1.0
FROM nvidia/cuda:11.8.0-devel-ubuntu22.04
ARG CUDA_VER=11.8
...