Schedule Visualization

About

visualize_lr_scheduler plots a learning rate schedule as an interactive Plotly chart. It helps you check a multiphase schedule before you train with it. It needs the d9d[visualization] extra.

Usage

Pass a function that builds your scheduler for a given optimizer. visualize_lr_scheduler builds the scheduler for a dummy optimizer, steps it num_steps times and plots the learning rate at each step.

import torch

from d9d.lr_scheduler.visualizer import visualize_lr_scheduler
from d9d.lr_scheduler.piecewise import piecewise_schedule, CurveLinear, CurveCosine


def create_scheduler(optimizer: torch.optim.Optimizer):
    return (
        piecewise_schedule(initial_multiplier=0.0, total_steps=100)
        .for_steps(10, 1.0, CurveLinear())
        .fill_rest(0.0, CurveCosine())
        .build(optimizer)
    )


# Opens an interactive plot in the browser or notebook.
visualize_lr_scheduler(
    factory=create_scheduler,
    num_steps=100,  # Number of steps to simulate
    init_lr=1e-3  # Base learning rate of the dummy optimizer
)

API Reference

`d9d.lr_scheduler.visualizer`

Plotting of learning rate schedules.

`visualize_lr_scheduler(factory, num_steps, init_lr=1.0)`

Plots a learning rate schedule as an interactive Plotly chart.

The function builds the scheduler for a dummy optimizer, steps it num_steps times and plots the learning rate at each step.

Parameters:

Name	Type	Description	Default
`factory`	`SchedulerFactory`	Callable that builds the scheduler for a given optimizer.	required
`num_steps`	`int`	Number of steps to simulate.	required
`init_lr`	`float`	Initial learning rate of the dummy optimizer.	`1.0`

Raises:

Type	Description
`ImportError`	If `plotly` is not installed.