You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
parser.add_argument('val', help='Validation data (.bin)')
193
197
parser.add_argument('--wdl', default=1.0, type=float, help='wdl=0.0 = train on evaluations, wdl=1.0 = train on game results, interpolates between (default=1.0)')
194
198
parser.add_argument('--batch-size', default=16384, type=int, help='Number of positions per batch / per iteration (default=16384)')
199
+
parser.add_argument('--train-size', type=int, help='Number of training samples to use (default=all)')
200
+
parser.add_argument('--val-size', type=int, help='Number of validation samples to use (default=all)')
195
201
parser.add_argument('--val-check-interval', default=2000, type=int, help='How often to check validation loss (default=2000)')
196
202
parser.add_argument('--log', action='store_true', help='Enable logging during training')
197
203
parser.add_argument('--top-n', default=3, type=int, help='Number of models to save for each epoch (default=3)')
0 commit comments