Neural machine translation with attention

This tutorial demonstrates how to train a sequence-to-sequence (seq2seq) model for Spanish-to-English translation roughly based on Effective Approaches to Attention-based Neural Machine Translation (Luong et al., 2015).

This tutorial: An encoder/decoder connected by attention.

While this architecture is somewhat outdated, it is still a very useful project to work through to get a deeper understanding of sequence-to-sequence models and attention mechanisms (before going on to Transformers).

This example assumes some knowledge of TensorFlow fundamentals below the level of a Keras layer:

• Working with tensors directly
• Writing custom keras.Models and keras.layers

After training the model in this notebook, you will be able to input a Spanish sentence, such as "¿todavia estan en casa?", and return the English translation: "are you still at home?"

The resulting model is exportable as a tf.saved_model, so it can be used in other TensorFlow environments.

The translation quality is reasonable for a toy example, but the generated attention plot is perhaps more interesting. This shows which parts of the input sentence has the model's attention while translating:

Note: This example takes approximately 10 minutes to run.

Setup

pip install -q -U tensorflow-text tf-keras
pip install -q einops

import os
os.environ['TF_USE_LEGACY_KERAS'] = '1'

import numpy as np

import typing
from typing import Any, Tuple

import einops
import matplotlib.pyplot as plt
import matplotlib.ticker as ticker

import tensorflow as tf
import tensorflow_text as tf_text

This tutorial uses a lot of low level API's where it's easy to get shapes wrong. This class is used to check shapes throughout the tutorial.

Toggle code

class ShapeChecker():
  def __init__(self):
    # Keep a cache of every axis-name seen
    self.shapes = {}

  def __call__(self, tensor, names, broadcast=False):
    if not tf.executing_eagerly():
      return

    parsed = einops.parse_shape(tensor, names)

    for name, new_dim in parsed.items():
      old_dim = self.shapes.get(name, None)

      if (broadcast and new_dim == 1):
        continue

      if old_dim is None:
        # If the axis name is new, add its length to the cache.
        self.shapes[name] = new_dim
        continue

      if new_dim != old_dim:
        raise ValueError(f"Shape mismatch for dimension: '{name}'\n"
                         f"    found: {new_dim}\n"
                         f"    expected: {old_dim}\n")

The data

The tutorial uses a language dataset provided by Anki. This dataset contains language translation pairs in the format:

May I borrow this book? ¿Puedo tomar prestado este libro?

They have a variety of languages available, but this example uses the English-Spanish dataset.

Download and prepare the dataset

For convenience, a copy of this dataset is hosted on Google Cloud, but you can also download your own copy. After downloading the dataset, here are the steps you need to take to prepare the data:

1. Add a start and end token to each sentence.
2. Clean the sentences by removing special characters.
3. Create a word index and reverse word index (dictionaries mapping from word → id and id → word).
4. Pad each sentence to a maximum length.

# Download the file
import pathlib

path_to_zip = tf.keras.utils.get_file(
    'spa-eng.zip', origin='http://storage.googleapis.com/download.tensorflow.org/data/spa-eng.zip',
    extract=True)

path_to_file = pathlib.Path(path_to_zip).parent/'spa-eng/spa.txt'
if not path_to_file.exists():
  path_to_file = pathlib.Path(path_to_zip)/'spa-eng/spa.txt'

Downloading data from http://storage.googleapis.com/download.tensorflow.org/data/spa-eng.zip
2638744/2638744 [==============================] - 0s 0us/step

def load_data(path):
  text = path.read_text(encoding='utf-8')

  lines = text.splitlines()
  pairs = [line.split('\t') for line in lines]

  context = np.array([context for target, context in pairs])
  target = np.array([target for target, context in pairs])

  return target, context

target_raw, context_raw = load_data(path_to_file)
print(context_raw[-1])

Si quieres sonar como un hablante nativo, debes estar dispuesto a practicar diciendo la misma frase una y otra vez de la misma manera en que un músico de banjo practica el mismo fraseo una y otra vez hasta que lo puedan tocar correctamente y en el tiempo esperado.

print(target_raw[-1])

If you want to sound like a native speaker, you must be willing to practice saying the same sentence over and over in the same way that banjo players practice the same phrase over and over until they can play it correctly and at the desired tempo.

Create a tf.data dataset

From these arrays of strings you can create a tf.data.Dataset of strings that shuffles and batches them efficiently:

BUFFER_SIZE = len(context_raw)
BATCH_SIZE = 64

is_train = np.random.uniform(size=(len(target_raw),)) < 0.8

train_raw = (
    tf.data.Dataset
    .from_tensor_slices((context_raw[is_train], target_raw[is_train]))
    .shuffle(BUFFER_SIZE)
    .batch(BATCH_SIZE))
val_raw = (
    tf.data.Dataset
    .from_tensor_slices((context_raw[~is_train], target_raw[~is_train]))
    .shuffle(BUFFER_SIZE)
    .batch(BATCH_SIZE))

for example_context_strings, example_target_strings in train_raw.take(1):
  print(example_context_strings[:5])
  print()
  print(example_target_strings[:5])
  break

tf.Tensor(
[b'Te dir\xc3\xa9 en cuanto lleguemos ah\xc3\xad.'
 b'El tiempo fue inmejorable.'
 b'Tom despert\xc3\xb3 a Mary sacudi\xc3\xa9ndola.'
 b'Tom le mand\xc3\xb3 unas flores a Mary.'
 b'El entrenador les dijo a los jugadores que no comieran mucho antes del partido.'], shape=(5,), dtype=string)

tf.Tensor(
[b"I'll tell you when we get there."
 b"The weather couldn't have been better." b'Tom shook Mary awake.'
 b'Tom sent Mary some flowers.'
 b'The coach told the players not to eat too much before the game.'], shape=(5,), dtype=string)

Text preprocessing

One of the goals of this tutorial is to build a model that can be exported as a tf.saved_model. To make that exported model useful it should take tf.string inputs, and return tf.string outputs: All the text processing happens inside the model. Mainly using a layers.TextVectorization layer.

Standardization

The model is dealing with multilingual text with a limited vocabulary. So it will be important to standardize the input text.

The first step is Unicode normalization to split accented characters and replace compatibility characters with their ASCII equivalents.

The tensorflow_text package contains a unicode normalize operation:

example_text = tf.constant('¿Todavía está en casa?')

print(example_text.numpy())
print(tf_text.normalize_utf8(example_text, 'NFKD').numpy())

b'\xc2\xbfTodav\xc3\xada est\xc3\xa1 en casa?'
b'\xc2\xbfTodavi\xcc\x81a esta\xcc\x81 en casa?'

Unicode normalization will be the first step in the text standardization function:

def tf_lower_and_split_punct(text):
  # Split accented characters.
  text = tf_text.normalize_utf8(text, 'NFKD')
  text = tf.strings.lower(text)
  # Keep space, a to z, and select punctuation.
  text = tf.strings.regex_replace(text, '[^ a-z.?!,¿]', '')
  # Add spaces around punctuation.
  text = tf.strings.regex_replace(text, '[.?!,¿]', r' \0 ')
  # Strip whitespace.
  text = tf.strings.strip(text)

  text = tf.strings.join(['[START]', text, '[END]'], separator=' ')
  return text

print(example_text.numpy().decode())
print(tf_lower_and_split_punct(example_text).numpy().decode())

¿Todavía está en casa?
[START] ¿ todavia esta en casa ? [END]

Text Vectorization

This standardization function will be wrapped up in a tf.keras.layers.TextVectorization layer which will handle the vocabulary extraction and conversion of input text to sequences of tokens.

max_vocab_size = 5000

context_text_processor = tf.keras.layers.TextVectorization(
    standardize=tf_lower_and_split_punct,
    max_tokens=max_vocab_size,
    ragged=True)

The TextVectorization layer and many other Keras preprocessing layers have an adapt method. This method reads one epoch of the training data, and works a lot like Model.fit. This adapt method initializes the layer based on the data. Here it determines the vocabulary:

context_text_processor.adapt(train_raw.map(lambda context, target: context))

# Here are the first 10 words from the vocabulary:
context_text_processor.get_vocabulary()[:10]

['',
 '[UNK]',
 np.str_('[START]'),
 np.str_('[END]'),
 np.str_('.'),
 np.str_('que'),
 np.str_('de'),
 np.str_('el'),
 np.str_('a'),
 np.str_('no')]

That's the Spanish TextVectorization layer, now build and .adapt() the English one:

target_text_processor = tf.keras.layers.TextVectorization(
    standardize=tf_lower_and_split_punct,
    max_tokens=max_vocab_size,
    ragged=True)

target_text_processor.adapt(train_raw.map(lambda context, target: target))
target_text_processor.get_vocabulary()[:10]

['',
 '[UNK]',
 np.str_('[START]'),
 np.str_('[END]'),
 np.str_('.'),
 np.str_('the'),
 np.str_('i'),
 np.str_('to'),
 np.str_('you'),
 np.str_('tom')]

Now these layers can convert a batch of strings into a batch of token IDs:

example_tokens = context_text_processor(example_context_strings)
example_tokens[:3, :]

<tf.RaggedTensor [[2, 30, 1795, 14, 151, 3841, 225, 4, 3], [2, 7, 60, 50, 1, 4, 3],
 [2, 10, 2209, 8, 32, 1, 4, 3]]>

The get_vocabulary method can be used to convert token IDs back to text:

context_vocab = np.array(context_text_processor.get_vocabulary())
tokens = context_vocab[example_tokens[0].numpy()]
' '.join(tokens)

'[START] te dire en cuanto lleguemos ahi . [END]'

The returned token IDs are zero-padded. This can easily be turned into a mask:

plt.subplot(1, 2, 1)
plt.pcolormesh(example_tokens.to_tensor())
plt.title('Token IDs')

plt.subplot(1, 2, 2)
plt.pcolormesh(example_tokens.to_tensor() != 0)
plt.title('Mask')

Text(0.5, 1.0, 'Mask')

Process the dataset

The process_text function below converts the Datasets of strings, into 0-padded tensors of token IDs. It also converts from a (context, target) pair to an ((context, target_in), target_out) pair for training with keras.Model.fit. Keras expects (inputs, labels) pairs, the inputs are the (context, target_in) and the labels are target_out. The difference between target_in and target_out is that they are shifted by one step relative to eachother, so that at each location the label is the next token.

def process_text(context, target):
  context = context_text_processor(context).to_tensor()
  target = target_text_processor(target)
  targ_in = target[:,:-1].to_tensor()
  targ_out = target[:,1:].to_tensor()
  return (context, targ_in), targ_out


train_ds = train_raw.map(process_text, tf.data.AUTOTUNE)
val_ds = val_raw.map(process_text, tf.data.AUTOTUNE)

Here is the first sequence of each, from the first batch:

for (ex_context_tok, ex_tar_in), ex_tar_out in train_ds.take(1):
  print(ex_context_tok[0, :10].numpy()) 
  print()
  print(ex_tar_in[0, :10].numpy()) 
  print(ex_tar_out[0, :10].numpy())

[   2   13  102 2745    7  905   19   21   67   12]

[  2  45   8  86 756  24   5 957  11   0]
[ 45   8  86 756  24   5 957  11   3   0]

The encoder/decoder

The following diagrams shows an overview of the model. In both the encoder is on the left, the decoder is on the right. At each time-step the decoder's output is combined with the encoder's output, to predict the next word.

The original [left] contains a few extra connections that are intentionally omitted from this tutorial's model [right], as they are generally unnecessary, and difficult to implement. Those missing connections are:

1. Feeding the state from the encoder's RNN to the decoder's RNN
2. Feeding the attention output back to the RNN's input.

The original from Effective Approaches to Attention-based Neural Machine Translation	This tutorial's model

Before getting into it define constants for the model:

UNITS = 256

The encoder

The goal of the encoder is to process the context sequence into a sequence of vectors that are useful for the decoder as it attempts to predict the next output for each timestep. Since the context sequence is constant, there is no restriction on how information can flow in the encoder, so use a bidirectional-RNN to do the processing:

A bidirectional RNN

The encoder:

1. Takes a list of token IDs (from context_text_processor).
2. Looks up an embedding vector for each token (Using a layers.Embedding).
3. Processes the embeddings into a new sequence (Using a bidirectional layers.GRU).
4. Returns the processed sequence. This will be passed to the attention head.

class Encoder(tf.keras.layers.Layer):
  def __init__(self, text_processor, units):
    super(Encoder, self).__init__()
    self.text_processor = text_processor
    self.vocab_size = text_processor.vocabulary_size()
    self.units = units

    # The embedding layer converts tokens to vectors
    self.embedding = tf.keras.layers.Embedding(self.vocab_size, units,
                                               mask_zero=True)

    # The RNN layer processes those vectors sequentially.
    self.rnn = tf.keras.layers.Bidirectional(
        merge_mode='sum',
        layer=tf.keras.layers.GRU(units,
                            # Return the sequence and state
                            return_sequences=True,
                            recurrent_initializer='glorot_uniform'))

  def call(self, x):
    shape_checker = ShapeChecker()
    shape_checker(x, 'batch s')

    # 2. The embedding layer looks up the embedding vector for each token.
    x = self.embedding(x)
    shape_checker(x, 'batch s units')

    # 3. The GRU processes the sequence of embeddings.
    x = self.rnn(x)
    shape_checker(x, 'batch s units')

    # 4. Returns the new sequence of embeddings.
    return x

  def convert_input(self, texts):
    texts = tf.convert_to_tensor(texts)
    if len(texts.shape) == 0:
      texts = tf.convert_to_tensor(texts)[tf.newaxis]
    context = self.text_processor(texts).to_tensor()
    context = self(context)
    return context

Try it out:

# Encode the input sequence.
encoder = Encoder(context_text_processor, UNITS)
ex_context = encoder(ex_context_tok)

print(f'Context tokens, shape (batch, s): {ex_context_tok.shape}')
print(f'Encoder output, shape (batch, s, units): {ex_context.shape}')

Context tokens, shape (batch, s): (64, 17)
Encoder output, shape (batch, s, units): (64, 17, 256)

The attention layer

The attention layer lets the decoder access the information extracted by the encoder. It computes a vector from the entire context sequence, and adds that to the decoder's output.

The simplest way you could calculate a single vector from the entire sequence would be to take the average across the sequence (layers.GlobalAveragePooling1D). An attention layer is similar, but calculates a weighted average across the context sequence. Where the weights are calculated from the combination of context and "query" vectors.

The attention layer

class CrossAttention(tf.keras.layers.Layer):
  def __init__(self, units, **kwargs):
    super().__init__()
    self.mha = tf.keras.layers.MultiHeadAttention(key_dim=units, num_heads=1, **kwargs)
    self.layernorm = tf.keras.layers.LayerNormalization()
    self.add = tf.keras.layers.Add()

  def call(self, x, context):
    shape_checker = ShapeChecker()

    shape_checker(x, 'batch t units')
    shape_checker(context, 'batch s units')

    attn_output, attn_scores = self.mha(
        query=x,
        value=context,
        return_attention_scores=True)

    shape_checker(x, 'batch t units')
    shape_checker(attn_scores, 'batch heads t s')

    # Cache the attention scores for plotting later.
    attn_scores = tf.reduce_mean(attn_scores, axis=1)
    shape_checker(attn_scores, 'batch t s')
    self.last_attention_weights = attn_scores

    x = self.add([x, attn_output])
    x = self.layernorm(x)

    return x

attention_layer = CrossAttention(UNITS)

# Attend to the encoded tokens
embed = tf.keras.layers.Embedding(target_text_processor.vocabulary_size(),
                                  output_dim=UNITS, mask_zero=True)
ex_tar_embed = embed(ex_tar_in)

result = attention_layer(ex_tar_embed, ex_context)

print(f'Context sequence, shape (batch, s, units): {ex_context.shape}')
print(f'Target sequence, shape (batch, t, units): {ex_tar_embed.shape}')
print(f'Attention result, shape (batch, t, units): {result.shape}')
print(f'Attention weights, shape (batch, t, s):    {attention_layer.last_attention_weights.shape}')

Context sequence, shape (batch, s, units): (64, 17, 256)
Target sequence, shape (batch, t, units): (64, 17, 256)
Attention result, shape (batch, t, units): (64, 17, 256)
Attention weights, shape (batch, t, s):    (64, 17, 17)

The attention weights will sum to 1 over the context sequence, at each location in the target sequence.

attention_layer.last_attention_weights[0].numpy().sum(axis=-1)

array([1.        , 1.        , 0.99999994, 1.        , 0.99999994,

       1.        , 1.        , 0.99999994, 1.        , 1.        ,
       1.        , 1.        , 1.        , 1.        , 1.        ,
       1.        , 1.        ], dtype=float32)

Here are the attention weights across the context sequences at t=0:

attention_weights = attention_layer.last_attention_weights
mask=(ex_context_tok != 0).numpy()

plt.subplot(1, 2, 1)
plt.pcolormesh(mask*attention_weights[:, 0, :])
plt.title('Attention weights')

plt.subplot(1, 2, 2)
plt.pcolormesh(mask)
plt.title('Mask');

Because of the small-random initialization the attention weights are initially all close to 1/(sequence_length). The model will learn to make these less uniform as training progresses.

The decoder

The decoder's job is to generate predictions for the next token at each location in the target sequence.

1. It looks up embeddings for each token in the target sequence.
2. It uses an RNN to process the target sequence, and keep track of what it has generated so far.
3. It uses RNN output as the "query" to the attention layer, when attending to the encoder's output.
4. At each location in the output it predicts the next token.

When training, the model predicts the next word at each location. So it's important that the information only flows in one direction through the model. The decoder uses a unidirectional (not bidirectional) RNN to process the target sequence.

When running inference with this model it produces one word at a time, and those are fed back into the model.

A unidirectional RNN

Here is the Decoder class' initializer. The initializer creates all the necessary layers.

class Decoder(tf.keras.layers.Layer):
  @classmethod
  def add_method(cls, fun):
    setattr(cls, fun.__name__, fun)
    return fun

  def __init__(self, text_processor, units):
    super(Decoder, self).__init__()
    self.text_processor = text_processor
    self.vocab_size = text_processor.vocabulary_size()
    self.word_to_id = tf.keras.layers.StringLookup(
        vocabulary=text_processor.get_vocabulary(),
        mask_token='', oov_token='[UNK]')
    self.id_to_word = tf.keras.layers.StringLookup(
        vocabulary=text_processor.get_vocabulary(),
        mask_token='', oov_token='[UNK]',
        invert=True)
    self.start_token = self.word_to_id('[START]')
    self.end_token = self.word_to_id('[END]')

    self.units = units


    # 1. The embedding layer converts token IDs to vectors
    self.embedding = tf.keras.layers.Embedding(self.vocab_size,
                                               units, mask_zero=True)

    # 2. The RNN keeps track of what's been generated so far.
    self.rnn = tf.keras.layers.GRU(units,
                                   return_sequences=True,
                                   return_state=True,
                                   recurrent_initializer='glorot_uniform')

    # 3. The RNN output will be the query for the attention layer.
    self.attention = CrossAttention(units)

    # 4. This fully connected layer produces the logits for each
    # output token.
    self.output_layer = tf.keras.layers.Dense(self.vocab_size)

Training

Next, the call method, takes 3 arguments:

• inputs - a context, x pair where:
  • context - is the context from the encoder's output.
  • x - is the target sequence input.
• state - Optional, the previous state output from the decoder (the internal state of the decoder's RNN). Pass the state from a previous run to continue generating text where you left off.
• return_state - [Default: False] - Set this to True to return the RNN state.

@Decoder.add_method
def call(self,
         context, x,
         state=None,
         return_state=False):  
  shape_checker = ShapeChecker()
  shape_checker(x, 'batch t')
  shape_checker(context, 'batch s units')

  # 1. Lookup the embeddings
  x = self.embedding(x)
  shape_checker(x, 'batch t units')

  # 2. Process the target sequence.
  x, state = self.rnn(x, initial_state=state)
  shape_checker(x, 'batch t units')

  # 3. Use the RNN output as the query for the attention over the context.
  x = self.attention(x, context)
  self.last_attention_weights = self.attention.last_attention_weights
  shape_checker(x, 'batch t units')
  shape_checker(self.last_attention_weights, 'batch t s')

  # Step 4. Generate logit predictions for the next token.
  logits = self.output_layer(x)
  shape_checker(logits, 'batch t target_vocab_size')

  if return_state:
    return logits, state
  else:
    return logits

That will be sufficient for training. Create an instance of the decoder to test out:

decoder = Decoder(target_text_processor, UNITS)

In training you'll use the decoder like this:

Given the context and target tokens, for each target token it predicts the next target token.

logits = decoder(ex_context, ex_tar_in)

print(f'encoder output shape: (batch, s, units) {ex_context.shape}')
print(f'input target tokens shape: (batch, t) {ex_tar_in.shape}')
print(f'logits shape shape: (batch, target_vocabulary_size) {logits.shape}')

encoder output shape: (batch, s, units) (64, 17, 256)
input target tokens shape: (batch, t) (64, 17)
logits shape shape: (batch, target_vocabulary_size) (64, 17, 5000)

Inference

To use it for inference you'll need a couple more methods:

@Decoder.add_method
def get_initial_state(self, context):
  batch_size = tf.shape(context)[0]
  start_tokens = tf.fill([batch_size, 1], self.start_token)
  done = tf.zeros([batch_size, 1], dtype=tf.bool)
  embedded = self.embedding(start_tokens)
  return start_tokens, done, self.rnn.get_initial_state(embedded)[0]

@Decoder.add_method
def tokens_to_text(self, tokens):
  words = self.id_to_word(tokens)
  result = tf.strings.reduce_join(words, axis=-1, separator=' ')
  result = tf.strings.regex_replace(result, '^ *\[START\] *', '')
  result = tf.strings.regex_replace(result, ' *\[END\] *$', '')
  return result

@Decoder.add_method
def get_next_token(self, context, next_token, done, state, temperature = 0.0):
  logits, state = self(
    context, next_token,
    state = state,
    return_state=True) 

  if temperature == 0.0:
    next_token = tf.argmax(logits, axis=-1)
  else:
    logits = logits[:, -1, :]/temperature
    next_token = tf.random.categorical(logits, num_samples=1)

  # If a sequence produces an `end_token`, set it `done`
  done = done | (next_token == self.end_token)
  # Once a sequence is done it only produces 0-padding.
  next_token = tf.where(done, tf.constant(0, dtype=tf.int64), next_token)

  return next_token, done, state

With those extra functions, you can write a generation loop:

# Setup the loop variables.
next_token, done, state = decoder.get_initial_state(ex_context)
tokens = []

for n in range(10):
  # Run one step.
  next_token, done, state = decoder.get_next_token(
      ex_context, next_token, done, state, temperature=1.0)
  # Add the token to the output.
  tokens.append(next_token)

# Stack all the tokens together.
tokens = tf.concat(tokens, axis=-1) # (batch, t)

# Convert the tokens back to a a string
result = decoder.tokens_to_text(tokens)
result[:3].numpy()

array([b'drunken stores emotional properly hold furniture cookies fortunate participate wed',
       b'organized tons objective misunderstanding sense minds toy uncle finally nasty',
       b'wondering autumn cold circumstances leaned candy praised shape cancel special'],
      dtype=object)

Since the model's untrained, it outputs items from the vocabulary almost uniformly at random.

The model

Now that you have all the model components, combine them to build the model for training:

class Translator(tf.keras.Model):
  @classmethod
  def add_method(cls, fun):
    setattr(cls, fun.__name__, fun)
    return fun

  def __init__(self, units,
               context_text_processor,
               target_text_processor):
    super().__init__()
    # Build the encoder and decoder
    encoder = Encoder(context_text_processor, units)
    decoder = Decoder(target_text_processor, units)

    self.encoder = encoder
    self.decoder = decoder

  def call(self, inputs):
    context, x = inputs
    context = self.encoder(context)
    logits = self.decoder(context, x)

    #TODO(b/250038731): remove this
    try:
      # Delete the keras mask, so keras doesn't scale the loss+accuracy. 
      del logits._keras_mask
    except AttributeError:
      pass

    return logits

During training the model will be used like this:

model = Translator(UNITS, context_text_processor, target_text_processor)

logits = model((ex_context_tok, ex_tar_in))

print(f'Context tokens, shape: (batch, s, units) {ex_context_tok.shape}')
print(f'Target tokens, shape: (batch, t) {ex_tar_in.shape}')
print(f'logits, shape: (batch, t, target_vocabulary_size) {logits.shape}')

Context tokens, shape: (batch, s, units) (64, 17)
Target tokens, shape: (batch, t) (64, 17)
logits, shape: (batch, t, target_vocabulary_size) (64, 17, 5000)

Train

For training, you'll want to implement your own masked loss and accuracy functions:

def masked_loss(y_true, y_pred):
    # Calculate the loss for each item in the batch.
    loss_fn = tf.keras.losses.SparseCategoricalCrossentropy(
        from_logits=True, reduction='none')
    loss = loss_fn(y_true, y_pred)

    # Mask off the losses on padding.
    mask = tf.cast(y_true != 0, loss.dtype)
    loss *= mask

    # Return the total.
    return tf.reduce_sum(loss)/tf.reduce_sum(mask)

def masked_acc(y_true, y_pred):
    # Calculate the loss for each item in the batch.
    y_pred = tf.argmax(y_pred, axis=-1)
    y_pred = tf.cast(y_pred, y_true.dtype)

    match = tf.cast(y_true == y_pred, tf.float32)
    mask = tf.cast(y_true != 0, tf.float32)

    return tf.reduce_sum(match)/tf.reduce_sum(mask)

Configure the model for training:

model.compile(optimizer='adam',
              loss=masked_loss, 
              metrics=[masked_acc, masked_loss])

The model is randomly initialized, and should give roughly uniform output probabilities. So it's easy to predict what the initial values of the metrics should be:

vocab_size = 1.0 * target_text_processor.vocabulary_size()

{"expected_loss": tf.math.log(vocab_size).numpy(),
 "expected_acc": 1/vocab_size}

{'expected_loss': np.float32(8.517193), 'expected_acc': 0.0002}

That should roughly match the values returned by running a few steps of evaluation:

model.evaluate(val_ds, steps=20, return_dict=True)

20/20 [==============================] - 7s 22ms/step - loss: 8.5299 - masked_acc: 3.5801e-04 - masked_loss: 8.5299
{'loss': 8.52992057800293,
 'masked_acc': 0.0003580127959139645,
 'masked_loss': 8.52992057800293}

history = model.fit(
    train_ds.repeat(), 
    epochs=100,
    steps_per_epoch = 100,
    validation_data=val_ds,
    validation_steps = 20,
    callbacks=[
        tf.keras.callbacks.EarlyStopping(patience=3)])

Epoch 1/100
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
W0000 00:00:1793452958.996772   99372 type_inference.cc:340] Type inference failed. This indicates an invalid graph that escaped type checking. Error message: INVALID_ARGUMENT: expected compatible input types, but input 1:
goo.gle/debugonly   
type_id: TFT_OPTIONAL
args {
  type_id: TFT_PRODUCT
  args {
    type_id: TFT_TENSOR
    args {
      type_id: TFT_INT32
    }
  }
}
 is neither a subtype nor a supertype of the combined inputs preceding it:
goo.gle/debugonly   
type_id: TFT_OPTIONAL
args {
  type_id: TFT_PRODUCT
  args {
    type_id: TFT_TENSOR
    args {
      type_id: TFT_INT8
    }
  }
}

    for Tuple type infernce function 0
    while inferring type of node 'cond_35/output/_22'
100/100 [==============================] - 23s 114ms/step - loss: 5.0437 - masked_acc: 0.2654 - masked_loss: 5.0437 - val_loss: 4.0794 - val_masked_acc: 0.3595 - val_masked_loss: 4.0794
Epoch 2/100
100/100 [==============================] - 6s 56ms/step - loss: 3.7001 - masked_acc: 0.4012 - masked_loss: 3.7001 - val_loss: 3.3776 - val_masked_acc: 0.4440 - val_masked_loss: 3.3776
Epoch 3/100
100/100 [==============================] - 4s 36ms/step - loss: 3.0884 - masked_acc: 0.4835 - masked_loss: 3.0884 - val_loss: 2.8437 - val_masked_acc: 0.5097 - val_masked_loss: 2.8437
Epoch 4/100
100/100 [==============================] - 4s 38ms/step - loss: 2.6936 - masked_acc: 0.5403 - masked_loss: 2.6936 - val_loss: 2.5085 - val_masked_acc: 0.5633 - val_masked_loss: 2.5085
Epoch 5/100
100/100 [==============================] - 4s 36ms/step - loss: 2.3903 - masked_acc: 0.5837 - masked_loss: 2.3903 - val_loss: 2.2223 - val_masked_acc: 0.6054 - val_masked_loss: 2.2223
Epoch 6/100
100/100 [==============================] - 3s 35ms/step - loss: 2.1740 - masked_acc: 0.6160 - masked_loss: 2.1740 - val_loss: 2.0364 - val_masked_acc: 0.6339 - val_masked_loss: 2.0364
Epoch 7/100
100/100 [==============================] - 3s 35ms/step - loss: 1.9935 - masked_acc: 0.6398 - masked_loss: 1.9935 - val_loss: 1.9007 - val_masked_acc: 0.6532 - val_masked_loss: 1.9007
Epoch 8/100
100/100 [==============================] - 3s 34ms/step - loss: 1.8741 - masked_acc: 0.6589 - masked_loss: 1.8741 - val_loss: 1.7817 - val_masked_acc: 0.6721 - val_masked_loss: 1.7817
Epoch 9/100
100/100 [==============================] - 3s 31ms/step - loss: 1.7823 - masked_acc: 0.6707 - masked_loss: 1.7823 - val_loss: 1.6777 - val_masked_acc: 0.6817 - val_masked_loss: 1.6777
Epoch 10/100
100/100 [==============================] - 3s 29ms/step - loss: 1.6962 - masked_acc: 0.6860 - masked_loss: 1.6962 - val_loss: 1.6627 - val_masked_acc: 0.6855 - val_masked_loss: 1.6627
Epoch 11/100
100/100 [==============================] - 3s 31ms/step - loss: 1.6325 - masked_acc: 0.6904 - masked_loss: 1.6325 - val_loss: 1.5586 - val_masked_acc: 0.6983 - val_masked_loss: 1.5586
Epoch 12/100
100/100 [==============================] - 3s 29ms/step - loss: 1.5559 - masked_acc: 0.7008 - masked_loss: 1.5559 - val_loss: 1.5137 - val_masked_acc: 0.7061 - val_masked_loss: 1.5137
Epoch 13/100
100/100 [==============================] - 3s 33ms/step - loss: 1.4729 - masked_acc: 0.7144 - masked_loss: 1.4729 - val_loss: 1.5402 - val_masked_acc: 0.7010 - val_masked_loss: 1.5402
Epoch 14/100
100/100 [==============================] - 3s 31ms/step - loss: 1.4686 - masked_acc: 0.7130 - masked_loss: 1.4686 - val_loss: 1.4626 - val_masked_acc: 0.7108 - val_masked_loss: 1.4626
Epoch 15/100
100/100 [==============================] - 3s 30ms/step - loss: 1.4058 - masked_acc: 0.7237 - masked_loss: 1.4066 - val_loss: 1.4636 - val_masked_acc: 0.7140 - val_masked_loss: 1.4636
Epoch 16/100
100/100 [==============================] - 3s 34ms/step - loss: 1.2049 - masked_acc: 0.7468 - masked_loss: 1.2049 - val_loss: 1.3535 - val_masked_acc: 0.7370 - val_masked_loss: 1.3535
Epoch 17/100
100/100 [==============================] - 3s 29ms/step - loss: 1.2137 - masked_acc: 0.7466 - masked_loss: 1.2137 - val_loss: 1.3335 - val_masked_acc: 0.7364 - val_masked_loss: 1.3335
Epoch 18/100
100/100 [==============================] - 3s 29ms/step - loss: 1.1852 - masked_acc: 0.7521 - masked_loss: 1.1852 - val_loss: 1.3641 - val_masked_acc: 0.7311 - val_masked_loss: 1.3641
Epoch 19/100
100/100 [==============================] - 3s 29ms/step - loss: 1.2043 - masked_acc: 0.7472 - masked_loss: 1.2043 - val_loss: 1.3477 - val_masked_acc: 0.7295 - val_masked_loss: 1.3477
Epoch 20/100
100/100 [==============================] - 3s 31ms/step - loss: 1.2179 - masked_acc: 0.7480 - masked_loss: 1.2179 - val_loss: 1.3481 - val_masked_acc: 0.7349 - val_masked_loss: 1.3481

plt.plot(history.history['loss'], label='loss')
plt.plot(history.history['val_loss'], label='val_loss')
plt.ylim([0, max(plt.ylim())])
plt.xlabel('Epoch #')
plt.ylabel('CE/token')
plt.legend()

<matplotlib.legend.Legend at 0x7f3c98498990>

plt.plot(history.history['masked_acc'], label='accuracy')
plt.plot(history.history['val_masked_acc'], label='val_accuracy')
plt.ylim([0, max(plt.ylim())])
plt.xlabel('Epoch #')
plt.ylabel('CE/token')
plt.legend()

<matplotlib.legend.Legend at 0x7f3c98422ed0>

Translate

Now that the model is trained, implement a function to execute the full text => text translation. This code is basically identical to the inference example in the decoder section, but this also captures the attention weights.

Toggle code

@Translator.add_method
def translate(self,
              texts, *,
              max_length=50,
              temperature=0.0):
  # Process the input texts
  context = self.encoder.convert_input(texts)
  batch_size = tf.shape(texts)[0]

  # Setup the loop inputs
  tokens = []
  attention_weights = []
  next_token, done, state = self.decoder.get_initial_state(context)

  for _ in range(max_length):
    # Generate the next token
    next_token, done, state = self.decoder.get_next_token(
        context, next_token, done,  state, temperature)

    # Collect the generated tokens
    tokens.append(next_token)
    attention_weights.append(self.decoder.last_attention_weights)

    if tf.executing_eagerly() and tf.reduce_all(done):
      break

  # Stack the lists of tokens and attention weights.
  tokens = tf.concat(tokens, axis=-1)   # t*[(batch 1)] -> (batch, t)
  self.last_attention_weights = tf.concat(attention_weights, axis=1)  # t*[(batch 1 s)] -> (batch, t s)

  result = self.decoder.tokens_to_text(tokens)
  return result

Here are the two helper methods, used above, to convert tokens to text, and to get the next token:

result = model.translate(['¿Todavía está en casa?']) # Are you still home
result[0].numpy().decode()

'is still at home ? '

Use that to generate the attention plot:

Toggle code

@Translator.add_method
def plot_attention(self, text, **kwargs):
  assert isinstance(text, str)
  output = self.translate([text], **kwargs)
  output = output[0].numpy().decode()

  attention = self.last_attention_weights[0]

  context = tf_lower_and_split_punct(text)
  context = context.numpy().decode().split()

  output = tf_lower_and_split_punct(output)
  output = output.numpy().decode().split()[1:]

  fig = plt.figure(figsize=(10, 10))
  ax = fig.add_subplot(1, 1, 1)

  ax.matshow(attention, cmap='viridis', vmin=0.0)

  fontdict = {'fontsize': 14}

  ax.set_xticklabels([''] + context, fontdict=fontdict, rotation=90)
  ax.set_yticklabels([''] + output, fontdict=fontdict)

  ax.xaxis.set_major_locator(ticker.MultipleLocator(1))
  ax.yaxis.set_major_locator(ticker.MultipleLocator(1))

  ax.set_xlabel('Input text')
  ax.set_ylabel('Output text')

model.plot_attention('¿Todavía está en casa?') # Are you still home

/tmpfs/tmp/ipykernel_99208/3355722706.py:23: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator. Otherwise, ticks may be mislabeled.
  ax.set_xticklabels([''] + context, fontdict=fontdict, rotation=90)
/tmpfs/tmp/ipykernel_99208/3355722706.py:24: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator. Otherwise, ticks may be mislabeled.
  ax.set_yticklabels([''] + output, fontdict=fontdict)

Translate a few more sentences and plot them:

%%time
# This is my life.
model.plot_attention('Esta es mi vida.')

CPU times: user 228 ms, sys: 5.9 ms, total: 234 ms
Wall time: 230 ms
/tmpfs/tmp/ipykernel_99208/3355722706.py:23: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator. Otherwise, ticks may be mislabeled.
  ax.set_xticklabels([''] + context, fontdict=fontdict, rotation=90)
/tmpfs/tmp/ipykernel_99208/3355722706.py:24: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator. Otherwise, ticks may be mislabeled.
  ax.set_yticklabels([''] + output, fontdict=fontdict)

%%time
 # Try to find out.'
model.plot_attention('Tratar de descubrir.')

CPU times: user 231 ms, sys: 3.66 ms, total: 235 ms
Wall time: 231 ms
/tmpfs/tmp/ipykernel_99208/3355722706.py:23: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator. Otherwise, ticks may be mislabeled.
  ax.set_xticklabels([''] + context, fontdict=fontdict, rotation=90)
/tmpfs/tmp/ipykernel_99208/3355722706.py:24: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator. Otherwise, ticks may be mislabeled.
  ax.set_yticklabels([''] + output, fontdict=fontdict)

The short sentences often work well, but if the input is too long the model literally loses focus and stops providing reasonable predictions. There are two main reasons for this:

1. The model was trained with teacher-forcing feeding the correct token at each step, regardless of the model's predictions. The model could be made more robust if it were sometimes fed its own predictions.
2. The model only has access to its previous output through the RNN state. If the RNN state looses track of where it was in the context sequence there's no way for the model to recover. Transformers improve on this by letting the decoder look at what it has output so far.

The raw data is sorted by length, so try translating the longest sequence:

long_text = context_raw[-1]

import textwrap
print('Expected output:\n', '\n'.join(textwrap.wrap(target_raw[-1])))

Expected output:
 If you want to sound like a native speaker, you must be willing to
practice saying the same sentence over and over in the same way that
banjo players practice the same phrase over and over until they can
play it correctly and at the desired tempo.

model.plot_attention(long_text)

/tmpfs/tmp/ipykernel_99208/3355722706.py:23: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator. Otherwise, ticks may be mislabeled.
  ax.set_xticklabels([''] + context, fontdict=fontdict, rotation=90)
/tmpfs/tmp/ipykernel_99208/3355722706.py:24: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator. Otherwise, ticks may be mislabeled.
  ax.set_yticklabels([''] + output, fontdict=fontdict)

The translate function works on batches, so if you have multiple texts to translate you can pass them all at once, which is much more efficient than translating them one at a time:

inputs = [
    'Hace mucho frio aqui.', # "It's really cold here."
    'Esta es mi vida.', # "This is my life."
    'Su cuarto es un desastre.' # "His room is a mess"
]

%%time
for t in inputs:
  print(model.translate([t])[0].numpy().decode())

print()

it is very cold here .
this is my life .
his room is a disaster . 

CPU times: user 700 ms, sys: 0 ns, total: 700 ms
Wall time: 684 ms

%%time
result = model.translate(inputs)

print(result[0].numpy().decode())
print(result[1].numpy().decode())
print(result[2].numpy().decode())
print()

it is very cold here . 
this is my life .  
his room is a disaster . 

CPU times: user 1.26 s, sys: 37.9 ms, total: 1.3 s
Wall time: 1.23 s

So overall this text generation function mostly gets the job done, but so you've only used it here in python with eager execution. Let's try to export it next:

Export

If you want to export this model you'll need to wrap the translate method in a tf.function. That implementation will get the job done:

class Export(tf.Module):
  def __init__(self, model):
    self.model = model

  @tf.function(input_signature=[tf.TensorSpec(dtype=tf.string, shape=[None])])
  def translate(self, inputs):
    return self.model.translate(inputs)

export = Export(model)

Run the tf.function once to compile it:

%%time
_ = export.translate(tf.constant(inputs))

W0000 00:00:1793453095.052535   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.053141   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.053699   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.054255   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.054805   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.055345   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.055907   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.056479   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.057054   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.057619   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.058163   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.058732   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.059512   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.060062   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.060605   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.061205   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.061850   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.062473   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.063032   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.063576   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.064120   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.064671   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.065211   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.065752   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.066302   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.066853   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.067388   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.067922   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.068466   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.069008   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.069540   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.070082   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.070642   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.071232   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.071858   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.072463   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.073005   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.073533   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.074073   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.074671   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.075233   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.075870   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.076484   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.077081   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.078114   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.078696   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.079228   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.079768   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.080300   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453095.080841   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
CPU times: user 58.1 s, sys: 4.75 s, total: 1min 2s
Wall time: 1min 2s

%%time
result = export.translate(tf.constant(inputs))

print(result[0].numpy().decode())
print(result[1].numpy().decode())
print(result[2].numpy().decode())
print()

it is very cold here .                                            
this is my life .                                             
his room is a disaster .                                            

CPU times: user 155 ms, sys: 14.8 ms, total: 169 ms
Wall time: 127 ms

Now that the function has been traced it can be exported using saved_model.save:

%%time
tf.saved_model.save(export, 'translator',
                    signatures={'serving_default': export.translate})

WARNING:tensorflow:Model's `__init__()` arguments contain non-serializable objects. Please implement a `get_config()` method in the subclassed Model for proper saving and loading. Defaulting to empty config.
WARNING:tensorflow:Model's `__init__()` arguments contain non-serializable objects. Please implement a `get_config()` method in the subclassed Model for proper saving and loading. Defaulting to empty config.
WARNING:tensorflow:Model's `__init__()` arguments contain non-serializable objects. Please implement a `get_config()` method in the subclassed Model for proper saving and loading. Defaulting to empty config.
WARNING:tensorflow:Model's `__init__()` arguments contain non-serializable objects. Please implement a `get_config()` method in the subclassed Model for proper saving and loading. Defaulting to empty config.
INFO:tensorflow:Assets written to: translator/assets
INFO:tensorflow:Assets written to: translator/assets
CPU times: user 1min 7s, sys: 4.88 s, total: 1min 12s
Wall time: 1min 12s

%%time
reloaded = tf.saved_model.load('translator')
_ = reloaded.translate(tf.constant(inputs)) #warmup

W0000 00:00:1793453178.035903   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453178.202725   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453178.257482   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453178.593339   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453181.803569   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453181.879530   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453182.185418   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453182.365181   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453182.374760   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453183.203634   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453183.213435   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453183.341752   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453183.438369   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453183.457634   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453183.492567   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453184.154832   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453184.164916   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453184.448785   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453184.530589   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453184.834284   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453184.846610   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453185.372313   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453185.383050   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453185.434616   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453185.444880   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453185.825681   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453186.143030   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453186.464480   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453186.530613   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453186.540646   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453186.644024   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453186.782507   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453186.843160   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453186.957523   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453187.064006   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453187.109230   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453187.119093   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453187.157474   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453187.573008   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453187.583691   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453187.709313   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453187.802449   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453187.881059   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453187.890767   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453188.249208   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453188.259795   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453188.345863   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453188.478719   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453188.512404   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453188.521790   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453188.567779   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453188.577297   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453189.134588   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453189.145915   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453189.470479   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453189.564573   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453189.781921   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453189.884745   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453190.317842   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453190.432211   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453190.475588   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453190.485650   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453190.679904   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453190.749529   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453190.765746   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453190.774972   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453190.797776   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453190.880144   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453190.890102   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453191.287384   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453191.612124   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453191.811809   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453191.821181   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.182259   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.312032   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.321833   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.339467   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.357763   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.368023   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.377471   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.418345   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.429500   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.439417   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.488533   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.514691   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.524156   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.534514   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.543895   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.588366   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.597802   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.626116   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.642426   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.651930   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.765823   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.775434   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.822110   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453192.832360   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.503607   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.514630   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.524220   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.534752   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.544174   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.560081   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.569659   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.585197   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.594949   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.611614   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.656334   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.665893   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.867606   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453194.879821   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453195.009096   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453195.201938   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453195.212111   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453195.358110   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453195.367428   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453195.496580   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453195.507277   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453195.725194   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453195.736624   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453195.960975   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453195.970541   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453196.350783   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453196.362727   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453196.421479   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453196.431897   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453196.642664   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453196.673362   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453196.683893   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453196.699927   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453196.709733   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453196.725372   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453197.741183   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453197.759124   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453210.708308   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.708913   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.709471   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.710034   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.710618   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.711168   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.711774   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.712415   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.713067   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.713705   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.714289   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.714843   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.715713   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.716295   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.716876   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.717419   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.717975   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.718538   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.719106   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.719650   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.720224   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.720809   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.721429   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.722024   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.722717   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.723361   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.723910   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.724456   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.725020   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.725572   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.726120   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.726729   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.727311   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.727895   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.728437   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.729010   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.729596   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.730175   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.730799   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.731374   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.731963   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.732617   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.733260   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.733874   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.735016   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.735599   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.736146   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.736695   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.737238   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
W0000 00:00:1793453210.737816   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
CPU times: user 46.4 s, sys: 1.98 s, total: 48.4 s
Wall time: 47.3 s

%%time
result = reloaded.translate(tf.constant(inputs))

print(result[0].numpy().decode())
print(result[1].numpy().decode())
print(result[2].numpy().decode())
print()

it is very cold here .                                            
this is my life .                                             
his room is a disaster .                                            

CPU times: user 165 ms, sys: 4.02 ms, total: 169 ms
Wall time: 125 ms

[Optional] Use a dynamic loop

It's worth noting that this initial implementation is not optimal. It uses a python loop:

for _ in range(max_length):
  ...
  if tf.executing_eagerly() and tf.reduce_all(done):
    break

The python loop is relatively simple but when tf.function converts this to a graph, it statically unrolls that loop. Unrolling the loop has two disadvantages:

1. It makes max_length copies of the loop body. So the generated graphs take longer to build, save and load.
2. You have to choose a fixed value for the max_length.
3. You can't break from a statically unrolled loop. The tf.function version will run the full max_length iterations on every call. That's why the break only works with eager execution. This is still marginally faster than eager execution, but not as fast as it could be.

To fix these shortcomings, the translate_dynamic method, below, uses a tensorflow loop:

for t in tf.range(max_length):
  ...
  if tf.reduce_all(done):
      break

It looks like a python loop, but when you use a tensor as the input to a for loop (or the condition of a while loop) tf.function converts it to a dynamic loop using operations like tf.while_loop.

There's no need for a max_length here it's just in case the model gets stuck generating a loop like: the united states of the united states of the united states....

On the down side, to accumulate tokens from this dynamic loop you can't just append them to a python list, you need to use a tf.TensorArray:

tokens = tf.TensorArray(tf.int64, size=1, dynamic_size=True)
...
for t in tf.range(max_length):
    ...
    tokens = tokens.write(t, next_token) # next_token shape is (batch, 1)
  ...
  tokens = tokens.stack()
  tokens = einops.rearrange(tokens, 't batch 1 -> batch t')

This version of the code can be quite a bit more efficient:

Toggle code

@Translator.add_method
def translate(self,
              texts,
              *,
              max_length=500,
              temperature=tf.constant(0.0)):
  shape_checker = ShapeChecker()
  context = self.encoder.convert_input(texts)
  batch_size = tf.shape(context)[0]
  shape_checker(context, 'batch s units')

  next_token, done, state = self.decoder.get_initial_state(context)

  # initialize the accumulator
  tokens = tf.TensorArray(tf.int64, size=1, dynamic_size=True)

  for t in tf.range(max_length):
    # Generate the next token
    next_token, done, state = self.decoder.get_next_token(
        context, next_token, done, state, temperature)
    shape_checker(next_token, 'batch t1')

    # Collect the generated tokens
    tokens = tokens.write(t, next_token)

    # if all the sequences are done, break
    if tf.reduce_all(done):
      break

  # Convert the list of generated token ids to a list of strings.
  tokens = tokens.stack()
  shape_checker(tokens, 't batch t1')
  tokens = einops.rearrange(tokens, 't batch 1 -> batch t')
  shape_checker(tokens, 'batch t')

  text = self.decoder.tokens_to_text(tokens)
  shape_checker(text, 'batch')

  return text

With eager execution this implementation performs on par with the original:

%%time
result = model.translate(inputs)

print(result[0].numpy().decode())
print(result[1].numpy().decode())
print(result[2].numpy().decode())
print()

it is very cold here . 
this is my life .  
his room is a disaster . 

CPU times: user 259 ms, sys: 7.35 ms, total: 266 ms
Wall time: 260 ms

But when you wrap it in a tf.function you'll notice two differences.

class Export(tf.Module):
  def __init__(self, model):
    self.model = model

  @tf.function(input_signature=[tf.TensorSpec(dtype=tf.string, shape=[None])])
  def translate(self, inputs):
    return self.model.translate(inputs)

export = Export(model)

First, it's much quicker to trace, since it only creates one copy of the loop body:

%%time
_ = export.translate(inputs)

W0000 00:00:1793453228.486317   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
CPU times: user 4.12 s, sys: 388 ms, total: 4.51 s
Wall time: 4.45 s

The tf.function is much faster than running with eager execution, and on small inputs it's often several times faster than the unrolled version, because it can break out of the loop.

%%time
result = export.translate(inputs)

print(result[0].numpy().decode())
print(result[1].numpy().decode())
print(result[2].numpy().decode())
print()

it is very cold here . 
this is my life .  
his room is a disaster . 

CPU times: user 28.3 ms, sys: 9.37 ms, total: 37.7 ms
Wall time: 24 ms

So save this version as well:

%%time
tf.saved_model.save(export, 'dynamic_translator',
                    signatures={'serving_default': export.translate})

WARNING:tensorflow:Model's `__init__()` arguments contain non-serializable objects. Please implement a `get_config()` method in the subclassed Model for proper saving and loading. Defaulting to empty config.
WARNING:tensorflow:Model's `__init__()` arguments contain non-serializable objects. Please implement a `get_config()` method in the subclassed Model for proper saving and loading. Defaulting to empty config.
WARNING:tensorflow:Model's `__init__()` arguments contain non-serializable objects. Please implement a `get_config()` method in the subclassed Model for proper saving and loading. Defaulting to empty config.
WARNING:tensorflow:Model's `__init__()` arguments contain non-serializable objects. Please implement a `get_config()` method in the subclassed Model for proper saving and loading. Defaulting to empty config.
INFO:tensorflow:Assets written to: dynamic_translator/assets
INFO:tensorflow:Assets written to: dynamic_translator/assets
CPU times: user 22.2 s, sys: 1.75 s, total: 24 s
Wall time: 23.9 s

%%time
reloaded = tf.saved_model.load('dynamic_translator')
_ = reloaded.translate(tf.constant(inputs)) #warmup

W0000 00:00:1793453257.805407   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453257.975084   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453258.253243   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453258.282864   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453258.464541   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453258.494892   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453258.583129   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453258.706423   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453258.747719   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453259.247281   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453259.411015   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453259.420859   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453259.868674   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453259.879134   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453259.929790   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453260.271566   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453260.282019   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453260.357460   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453260.367407   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453260.596495   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453260.613302   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453260.623101   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453260.865800   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453260.875976   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453261.151899   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453261.705140   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453261.716172   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453261.725904   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453261.933007   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453261.943276   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453261.975660   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453261.985306   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453262.134117   99208 graph_constructor.cc:846] Node 'cond/while' has 13 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453262.145062   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 40 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453262.317387   99208 graph_constructor.cc:846] Node 'cond/while' has 14 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453262.327563   99208 graph_constructor.cc:846] Node 'cond' has 4 outputs but the _output_shapes attribute specifies shapes for 42 outputs. Output shapes may be inaccurate.
W0000 00:00:1793453264.085994   99208 op_level_cost_estimator.cc:699] Error in PredictCost() for the op: goo.gle/debugonly   op: "Softmax" attr { key: "T" value { type: DT_FLOAT } } inputs { dtype: DT_FLOAT shape { unknown_rank: true } } device { type: "GPU" vendor: "NVIDIA" model: "Tesla T4" frequency: 1590 num_cores: 40 environment { key: "architecture" value: "7.5" } environment { key: "cuda" value: "12050" } environment { key: "cudnn" value: "90300" } num_registers: 65536 l1_cache_size: 24576 l2_cache_size: 4194304 shared_memory_size_per_multiprocessor: 65536 memory_size: 14301462528 bandwidth: 320064000 } outputs { dtype: DT_FLOAT shape { unknown_rank: true } }
CPU times: user 11.3 s, sys: 749 ms, total: 12.1 s
Wall time: 12 s

%%time
result = reloaded.translate(tf.constant(inputs))

print(result[0].numpy().decode())
print(result[1].numpy().decode())
print(result[2].numpy().decode())
print()

it is very cold here . 
this is my life .  
his room is a disaster . 

CPU times: user 38.7 ms, sys: 1.29 ms, total: 39.9 ms
Wall time: 25 ms

Next steps

• Download a different dataset to experiment with translations, for example, English to German, or English to French.
• Experiment with training on a larger dataset, or using more epochs.
• Try the transformer tutorial which implements a similar translation task but uses transformer layers instead of RNNs. This version also uses a text.BertTokenizer to implement word-piece tokenization.
• Visit the tensorflow_addons.seq2seq tutorial, which demonstrates a higher-level functionality for implementing this sort of sequence-to-sequence model, such as seq2seq.BeamSearchDecoder.