""" Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved. SPDX-License-Identifier: MIT-0 Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. """ import json from transformers import AutoTokenizer, AutoModelForQuestionAnswering from aws_lambda_powertools import Metrics, Logger, Tracer from aws_lambda_powertools.metrics import MetricUnit import os import boto3 import torch # Lambda Powertools Setup logger = Logger() tracer = Tracer() metrics = Metrics() # Use tmp directory for model storage from S3. MODEL_NLP_DIR = ("/tmp/nlp1/") s3 = boto3.resource('s3') # Downloads the correct model from Amazon S3 @tracer.capture_method def util_dl_s3_model(): logger.info('Download model files from S3') if not os.path.isdir(MODEL_NLP_DIR): os.mkdir(MODEL_NLP_DIR) logger.info('Created folder: ' + MODEL_NLP_DIR) else: logger.info('Folder already exists: ' + MODEL_NLP_DIR) s3.meta.client.download_file(os.environ['S3_MODEL_BUCKET_NAME'], os.environ['S3_NLP1_MODEL'], '/tmp/nlp1/pytorch_model.bin') s3.meta.client.download_file(os.environ['S3_MODEL_BUCKET_NAME'], os.environ['S3_NLP1_CONFIG'], '/tmp/nlp1/config.json') s3.meta.client.download_file(os.environ['S3_MODEL_BUCKET_NAME'], os.environ['S3_NLP1_TOKENIZER'], '/tmp/nlp1/tokenizer.json') s3.meta.client.download_file(os.environ['S3_MODEL_BUCKET_NAME'], os.environ['S3_NLP1_TOKENIZER_CONFIG'], '/tmp/nlp1/tokenizer_config.json') tokenizer = AutoTokenizer.from_pretrained("/tmp/nlp1/") model = AutoModelForQuestionAnswering.from_pretrained("/tmp/nlp1/") return [model, tokenizer] # Load the model outside of handler (Single Model Inference) loaded_model_tokenizer = util_dl_s3_model() @tracer.capture_lambda_handler @metrics.log_metrics(capture_cold_start_metric=True) @logger.inject_lambda_context(log_event=True) def lambda_handler(event, context): body = json.loads(event['body']) question = body['question'] context = body['context'] # Gather the inputs inputs = loaded_model_tokenizer[1].encode_plus(question,context,add_special_tokens=True,return_tensors="pt") input_ids = inputs["input_ids"].tolist()[0] # Perform the inference output = loaded_model_tokenizer[0](**inputs) answer_start_scores = output.start_logits answer_end_scores = output.end_logits answer_start = torch.argmax(answer_start_scores) answer_end = torch.argmax(answer_end_scores) + 1 answer = loaded_model_tokenizer[1].convert_tokens_to_string(loaded_model_tokenizer[1].convert_ids_to_tokens(input_ids[answer_start:answer_end])) print('Question: {0}, Answer: {1}'.format(question, answer)) return { 'statusCode': 200, 'headers': {'Content-Type': 'application/json'}, 'body': json.dumps({ 'Question': question, 'Answer': answer }) }