WandB¶
plotify.wandb_plots.fetch_smooth_curves(x_key, y_key, wandb_ids, samples = 2048, max_x = None, smooth_temperature = 0.0)
¶
plotify.wandb_plots.average_align_curves(x_curves, y_curves, samples = 2048, x_scale = 'linear')
¶
plotify.wandb_plots.wandb_plot(config)
¶
Generates a new plot from a configuration dictionary.
Example configuration:
config = {
'type': pl.PublicationPlot,
'title': 'PT Tasks',
'subtitle': 'mini-ImageNet',
'xtitle': 'Iterations',
'ytitle': 'Accuracy',
'xlims': (0, 1000),
'ylims': (0.0, 1.0),
'legend': {
'inset': True,
'loc': 'best',
'show': True,
},
'results': [ # Each dict is a curve
{
'wandb_id': [
'arnolds/qmcrl/51podsf0',
'arnolds/qmcrl/xpouxt8w',
],
'x_key': 'iteration',
'y_key': 'test/episode_returns',
'label': 'MC',
'color': pl.Maureen['blue'],
'linewidth': 1.8,
'smooth_window': 1,
'markevery': 1000,
'samples': 4196,
'shade': 'std',
},
{
'wandb_id': 'arnolds/meta-features/11r23eby',
'x_key': 'iteration',
'y_key': 'valid/accuracy',
'label': 'No Varnish',
'color': pl.Maureen['orange'],
'temperature': 50.0,
'markevery': 100,
},
],
}