Initial push of gensim_with_word2vec_model_artifacts_local_serving sample code

Author: eitansela

.gitignore

@@ -88,3 +88,7 @@ scikit_learn_bring_your_own_container_and_own_model_local_serving/data/train/cal
 scikit_learn_bring_your_own_container_and_own_model_local_serving/data/validation/california_validation.csv
 scikit_learn_bring_your_own_container_and_own_model_local_serving/model.joblib
 scikit_learn_bring_your_own_container_and_own_model_local_serving/model.tar.gz
+gensim_with_word2vec_model_artifacts_local_serving/eval.json
+gensim_with_word2vec_model_artifacts_local_serving/model.tar.gz
+gensim_with_word2vec_model_artifacts_local_serving/vectors.bin
+gensim_with_word2vec_model_artifacts_local_serving/vectors.txt

gensim_with_word2vec_model_artifacts_local_serving/code/inference.py

@@ -0,0 +1,29 @@
+import os
+import json
+from gensim.models import KeyedVectors
+
+def input_fn(request_body, request_content_type):
+ print(f"request_body: {request_body}")
+ print(type(request_body))
+ if request_content_type == "application/json":
+ payload = json.loads(request_body)
+ instances = payload["instances"]
+ return instances
+
+
+def predict_fn(instances, word_vectors):
+ #########################################
+ # Do your custom preprocessing logic here
+ #########################################
+
+ print(f"instances: {instances}")
+ print("calling model")
+ predictions = word_vectors.most_similar(positive=instances)
+ return predictions
+
+
+def model_fn(model_dir):
+ print("loading model from: {}".format(model_dir))
+ word_vectors = KeyedVectors.load_word2vec_format(os.path.join(model_dir, "vectors.txt"), binary=False)
+ print(f'word vectors length: {len(word_vectors)}')
+ return word_vectors

gensim_with_word2vec_model_artifacts_local_serving/code/requirements.txt

@@ -0,0 +1 @@
+gensim==4.1.2

gensim_with_word2vec_model_artifacts_local_serving/gensim_with_word2vec_model_artifacts_local_serving.py

@@ -0,0 +1,55 @@
+# This is a sample Python program that serve a Word2Vec model, trained with BlazingText algorithm with inference using gensim.
+# This implementation will work on your *local computer* or in the *AWS Cloud*.
+#
+# Prerequisites:
+# 1. Install required Python packages:
+# `pip install -r requirements.txt`
+# 2. Docker Desktop installed and running on your computer:
+# `docker ps`
+# 3. You should have AWS credentials configured on your local machine
+# in order to be able to pull the docker image from ECR.
+###############################################################################################
+
+import boto3
+from sagemaker.deserializers import JSONDeserializer
+from sagemaker.serializers import JSONSerializer
+from sagemaker.sklearn import SKLearnModel
+
+
+DUMMY_IAM_ROLE = 'arn:aws:iam::111111111111:role/service-role/AmazonSageMaker-ExecutionRole-20200101T000001'
+s3 = boto3.client('s3')
+
+
+def main():
+
+ # Download a pre-trained model archive file
+ print('Downloading a pre-trained model archive file')
+ s3.download_file('aws-ml-blog', 'artifacts/word2vec_algorithm_model_artifacts/model.tar.gz', 'model.tar.gz')
+
+ print('Deploying endpoint in local mode')
+ print(
+ 'Note: if launching for the first time in local mode, container image download might take a few minutes to complete.')
+ model = SKLearnModel(
+ role=DUMMY_IAM_ROLE,
+ model_data='file://./model.tar.gz',
+ framework_version='0.23-1',
+ py_version='py3',
+ source_dir='code',
+ entry_point='inference.py'
+ )
+
+ print('Deploying endpoint in local mode')
+ predictor = model.deploy(initial_instance_count=1, instance_type='local')
+
+ payload = {"instances": ["king","queen"]}
+ predictor.serializer = JSONSerializer()
+ predictor.deserializer = JSONDeserializer()
+ predictions = predictor.predict(payload)
+ print(f"Predictions: {predictions}")
+
+ print('About to delete the endpoint to stop paying (if in cloud mode).')
+ predictor.delete_endpoint(predictor.endpoint_name)
+
+
+if __name__ == "__main__":
+ main()

gensim_with_word2vec_model_artifacts_local_serving/requirements.txt

@@ -0,0 +1,4 @@
+numpy
+pandas
+sagemaker>=2.0.0<3.0.0
+sagemaker[local]