Path: blob/master/Natural Language Processing with Attention Models/Week 1 - Neural Machine Translation/output_dir/train/events.out.tfevents.1608282144.fb630eec7870
18519 views
�K" ��A
brain.Event:2Wl�. ��W� #���A*!
metrics/CrossEntropyLossC�&A����s ̮�P �J��A*�
�
gin_configB�B�#### Parameters for Adam:
Adam.b1 = 0.9
Adam.b2 = 0.999
Adam.clip_grad_norm = None
Adam.eps = 1e-05
Adam.weight_decay_rate = 1e-05
#### Parameters for AddLossWeights:
# None.
#### Parameters for backend:
backend.name = 'jax'
#### Parameters for BucketByLength:
BucketByLength.length_axis = 0
BucketByLength.strict_pad_on_len = False
#### Parameters for FilterByLength:
FilterByLength.length_axis = 0
#### Parameters for LogSoftmax:
LogSoftmax.axis = -1
#### Parameters for random_spans_helper:
# None.
#### Parameters for SentencePieceVocabulary:
# None.
#### Parameters for data.TFDS:
# None.
#### Parameters for tf_inputs.TFDS:
# None.
#### Parameters for data.Tokenize:
# None.
#### Parameters for tf_inputs.Tokenize:
tf_inputs.Tokenize.keys = None
tf_inputs.Tokenize.n_reserved_ids = 0
tf_inputs.Tokenize.vocab_type = 'subword'
#### Parameters for Vocabulary:
# None.
#### Parameters for warmup_and_rsqrt_decay:
# None.J
text1��K, ���E �W��A*
training/learning_rateĚ'7��ڜ/ m]P �X��A*"
training/steps per secondl�<C��+ ��K �Y��A*
training/gradients_l28��?�[y�# ��wC \Z��A*
training/lossC�&A��,) 7�_ �Z��A*
training/weights_l2��E�bH. ��W� �5�i��A
*!
metrics/CrossEntropyLosss$A�~;, ���E C�i��A
*
training/learning_rateu��8��g�/ m]P �C�i��A
*"
training/steps per second���<���+ ��K �D�i��A
*
training/gradients_l2/�@XE1# ��wC GE�i��A
*
training/losss$A�AL]) 7�_ �E�i��A
*
training/weights_l2$��Es��g