sdks/python/apache_beam/examples/inference/onnx_sentiment_classification.py

#
# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements.  See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to You under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License.  You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#

""""A pipeline that uses RunInference to perform sentiment classification
using RoBERTa.

This pipeline takes sentences from a custom text file, and then uses RoBERTa
from Hugging Face to predict the sentiment of a given review. The pipeline
then writes the prediction to an output file in which users can then compare against true labels.

Model is fine-tuned RoBERTa from
https://github.com/SeldonIO/seldon-models/blob/master/pytorch/moviesentiment_roberta/pytorch-roberta-onnx.ipynb # pylint: disable=line-too-long
"""

import argparse
import logging
from typing import Iterable
from typing import Iterator
from typing import Tuple

import numpy as np

import apache_beam as beam
import torch
from apache_beam.ml.inference.base import KeyedModelHandler
from apache_beam.ml.inference.base import PredictionResult
from apache_beam.ml.inference.base import RunInference
from apache_beam.ml.inference.onnx_inference import OnnxModelHandlerNumpy
from apache_beam.options.pipeline_options import PipelineOptions
from apache_beam.options.pipeline_options import SetupOptions
from apache_beam.runners.runner import PipelineResult
from transformers import RobertaTokenizer


def tokenize_sentence(text: str,
                      tokenizer: RobertaTokenizer) -> Tuple[str, torch.Tensor]:
  tokenized_sentence = tokenizer.encode(text, add_special_tokens=True)

  # Workaround to manually remove batch dim until we have the feature to
  # add optional batching flag.
  # TODO(https://github.com/apache/beam/issues/21863): Remove once optional
  # batching flag added
  return text, torch.tensor(tokenized_sentence).numpy()


def filter_empty_lines(text: str) -> Iterator[str]:
  if len(text.strip()) > 0:
    yield text


class PostProcessor(beam.DoFn):
  def process(self, element: Tuple[str, PredictionResult]) -> Iterable[str]:
    filename, prediction_result = element
    prediction = np.argmax(prediction_result.inference, axis=0)
    yield filename + ';' + str(prediction)


def parse_known_args(argv):
  """Parses args for the workflow."""
  parser = argparse.ArgumentParser()
  parser.add_argument(
      '--input',
      dest='input',
      help='Path to the text file containing sentences.')
  parser.add_argument(
      '--output',
      dest='output',
      required=True,
      help='Path of file in which to save the output predictions.')
  parser.add_argument(
      '--model_uri',
      dest='model_uri',
      required=True,
      help="Path to the model's uri.")
  return parser.parse_known_args(argv)


def run(
    argv=None, save_main_session=True, test_pipeline=None) -> PipelineResult:
  """
  Args:
    argv: Command line arguments defined for this example.
    save_main_session: Used for internal testing.
    test_pipeline: Used for internal testing.
  """
  known_args, pipeline_args = parse_known_args(argv)
  pipeline_options = PipelineOptions(pipeline_args)
  pipeline_options.view_as(SetupOptions).save_main_session = save_main_session

  # TODO: Remove once nested tensors https://github.com/pytorch/nestedtensor
  # is officially released.
  class OnnxNoBatchModelHandler(OnnxModelHandlerNumpy):
    """Wrapper to OnnxModelHandlerNumpy to limit batch size to 1.

    The tokenized strings generated from RobertaTokenizer may have different
    lengths, which doesn't work with torch.stack() in current RunInference
    implementation since stack() requires tensors to be the same size.

    Restricting max_batch_size to 1 means there is only 1 example per `batch`
    in the run_inference() call.
    """
    def batch_elements_kwargs(self):
      return {'max_batch_size': 1}

  model_handler = OnnxNoBatchModelHandler(model_uri=known_args.model_uri)

  pipeline = test_pipeline
  if not test_pipeline:
    pipeline = beam.Pipeline(options=pipeline_options)

  tokenizer = RobertaTokenizer.from_pretrained('roberta-base')

  text = (pipeline | 'ReadSentences' >> beam.io.ReadFromText(known_args.input))
  text_and_tokenized_text_tuple = (
      text
      | 'FilterEmptyLines' >> beam.ParDo(filter_empty_lines)
      |
      'TokenizeSentence' >> beam.Map(lambda x: tokenize_sentence(x, tokenizer)))
  output = (
      text_and_tokenized_text_tuple
      | 'PyTorchRunInference' >> RunInference(KeyedModelHandler(model_handler))
      | 'ProcessOutput' >> beam.ParDo(PostProcessor()))
  _ = output | "WriteOutput" >> beam.io.WriteToText(
      known_args.output, shard_name_template='', append_trailing_newlines=True)

  result = pipeline.run()
  result.wait_until_finish()
  return result


if __name__ == '__main__':
  logging.getLogger().setLevel(logging.INFO)
  run()