def serve(model, port=16118, debug=True, **fit_kwargs):
"""
:param model: Sklearn-compatible model, that is pickleable and has interface for .fit(X, y) and .predict_proba(X)
:param port: specify localhost port where the model will be served
:param debug: whether to start at debug mode
:return:
"""
init_model_server(create_model_func=partial(SklearnTextClassifier,
model=model),
train_script=partial(fit_sklearn_classifier,
model=model),
redis_queue=os.environ.get('RQ_QUEUE_NAME', 'default'),
redis_host=os.environ.get('REDIS_HOST', 'localhost'),
redis_port=os.environ.get('REDIS_HOST', 6379),
**fit_kwargs)
app.run(host='localhost', port=port, debug=debug)
def serve(learner_script,
port=16118,
debug=True,
image_dir='~/.heartex/images',
num_iter=10,
**fit_kwargs):
"""
:param learner_script: function that takes (DataBunch, str) as input and returns Learner object
:param port: specify localhost port where the model will be served
:param debug: whether to start at debug mode
:return:
"""
init_model_server(create_model_func=FastaiImageClassifier,
train_script=fit_fastai_image_classifier,
learner_script=learner_script,
num_iter=num_iter,
image_dir=image_dir,
redis_queue=os.environ.get('RQ_QUEUE_NAME', 'default'),
redis_host=os.environ.get('REDIS_HOST', 'localhost'),
redis_port=os.environ.get('REDIS_HOST', 6379),
**fit_kwargs)
app.run(host='0.0.0.0', port=port, debug=debug)
import os
import argparse
from ner import TransformersBasedTagger, train_ner
from htx import app, init_model_server
init_model_server(
# refer to class definition that inherits htx.BaseModel and implements load() and predict() methods
create_model_func=TransformersBasedTagger,
# training script that will be called within RQ
train_script=train_ner,
# name of the Redis queue
redis_queue=os.environ.get('RQ_QUEUE_NAME', 'default'),
# Redis host
redis_host=os.environ.get('REDIS_HOST', 'localhost'),
# here we pass the kwargs parameters to train script
pretrained_model=os.environ.get('pretrained_model', 'bert-base-uncased'),
cache_dir=os.environ.get('cache_dir', '~/.heartex/cache'))
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument('--port', dest='port', default='9090')
args = parser.parse_args()
app.run(host='localhost', port=args.port, debug=True)
import os
from htx import app, init_model_server
from bert_classifier.train import train_classifier
from bert_classifier.serve import LabelStudioTransformersClassifier
init_model_server(
create_model_func=LabelStudioTransformersClassifier,
train_script=train_classifier,
# name of the Redis queue
redis_queue=os.environ.get('RQ_QUEUE_NAME', 'default'),
# Redis host
redis_host=os.environ.get('REDIS_HOST', 'localhost'),
# here we pass the kwargs parameters to train script
pretrained_model=os.environ.get('pretrained_model',
'bert-base-multilingual-cased'),
cache_dir=os.environ.get('cache_dir', '/data/cache'),
model_dir=os.environ.get('model_dir', '/data/model'),
train_logs=os.environ.get('train_logs', '/data/train_logs'))
import os
import logging
import argparse
logging.basicConfig(level=logging.INFO)
from htx import app, init_model_server
from image_classifier import FastaiImageClassifier, train_script
init_model_server(
create_model_func=FastaiImageClassifier,
train_script=train_script,
num_iter=10,
image_dir='~/.heartex/images',
redis_queue=os.environ.get('RQ_QUEUE_NAME', 'default'),
redis_host=os.environ.get('REDIS_HOST', 'localhost'),
redis_port=os.environ.get('REDIS_PORT', 6379),
)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument('--port', dest='port', default='9090')
args = parser.parse_args()
app.run(host='localhost', port=args.port, debug=True)
import os
import argparse
from tfhub_classifier import TFHubClassifier
from htx import app, init_model_server
init_model_server(
# refer to class definition that inherits htx.BaseModel and implements load() and predict() methods
create_model_func=TFHubClassifier,
# training script that will be called within RQ
train_script=TFHubClassifier.fit_single_label,
# name of the Redis queue
redis_queue=os.environ.get('RQ_QUEUE_NAME', 'default'),
# Redis host
redis_host=os.environ.get('REDIS_HOST', 'localhost'),
# here we pass the kwargs parameters to train script
tfhub_module_spec=os.environ.get(
'tfhub_module_spec',
'https://tfhub.dev/google/universal-sentence-encoder/2'))
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument('--port', dest='port', default='9090')
args = parser.parse_args()
app.run(host='localhost', port=args.port, debug=True)
import os
import logging
import argparse
logging.basicConfig(level=logging.INFO)
from htx import app, init_model_server
from image_classifier import FastaiImageClassifier, train_script
from settings import REDIS_HOST, REDIS_PORT, RQ_QUEUE_NAME
init_model_server(
create_model_func=FastaiImageClassifier,
train_script=train_script,
num_iter=10,
image_dir='~/.heartex/images',
redis_queue=RQ_QUEUE_NAME,
redis_host=REDIS_HOST,
redis_port=REDIS_PORT,
)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument('--port', dest='port', default='9090')
args = parser.parse_args()
app.run(host='localhost', port=args.port, debug=True)