Created
September 4, 2023 15:04
-
-
Save pythonlessons/d9b05217f2b38e870840880439015bc5 to your computer and use it in GitHub Desktop.
transformers_training
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| # Define callbacks | |
| warmupCosineDecay = WarmupCosineDecay( | |
| lr_after_warmup=configs.lr_after_warmup, | |
| final_lr=configs.final_lr, | |
| warmup_epochs=configs.warmup_epochs, | |
| decay_epochs=configs.decay_epochs, | |
| initial_lr=configs.init_lr, | |
| ) | |
| earlystopper = EarlyStopping(monitor="val_masked_accuracy", patience=5, verbose=1, mode="max") | |
| checkpoint = ModelCheckpoint(f"{configs.model_path}/model.h5", monitor="val_masked_accuracy", verbose=1, save_best_only=True, mode="max", save_weights_only=False) | |
| tb_callback = TensorBoard(f"{configs.model_path}/logs") | |
| reduceLROnPlat = ReduceLROnPlateau(monitor="val_masked_accuracy", factor=0.9, min_delta=1e-10, patience=2, verbose=1, mode="max") | |
| model2onnx = Model2onnx(f"{configs.model_path}/model.h5", metadata={"tokenizer": tokenizer.dict(), "detokenizer": detokenizer.dict()}, save_on_epoch_end=False) | |
| encDecSplitCallback = EncDecSplitCallback(configs.model_path, encoder_metadata={"tokenizer": tokenizer.dict()}, decoder_metadata={"detokenizer": detokenizer.dict()}) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment