Ray

Ray is a distributed computing framework specifically designed for large-scale parallel tasks and reinforcement learning applications. It was developed by a research team from the University of California, Berkeley, aiming to simplify the process of building high-performance, scalable distributed applications. Ray supports Python and Java, and can be easily integrated into existing machine learning, data processing, and reinforcement learning workflows.

SwanLab supports Ray experiment logging. You can conveniently record experiment metrics and hyperparameters through SwanLabLoggerCallback.

1. Import SwanLabCallback

from swanlab.integration.ray import SwanLabLoggerCallback

SwanLabLoggerCallback is a log recording class adapted for Ray.

The parameters that can be defined in SwanLabLoggerCallback include:

• project: Project name
• workspace: Workspace name
• Other parameters consistent with swanlab.init

2. Integration with tune.Tuner

tuner = tune.Tuner(
    ...
    run_config=tune.RunConfig(
        callbacks=[SwanLabLoggerCallback(project="Ray_Project")],
    ),
)

3. Complete Example

import random
from ray import tune
from swanlab.integration.ray import SwanLabLoggerCallback

def train_func(config):
    offset = random.random() / 5
    for epoch in range(2, config["epochs"]):
        acc = 1 - (2 + config["lr"]) ** -epoch - random.random() / epoch - offset
        loss = (2 + config["lr"]) ** -epoch + random.random() / epoch + offset
        tune.report({"acc": acc, "loss": loss})


tuner = tune.Tuner(
    train_func,
    param_space={
        "lr": tune.grid_search([0.001, 0.01, 0.1, 1.0]),
        "epochs": 10,
    },
    run_config=tune.RunConfig(
        callbacks=[SwanLabLoggerCallback(project="Ray_Project")],
    ),
)
results = tuner.fit()