import wandb
import json
from text_classification import config, data, predict, utils
app = FastAPI(
title="text-classification",
description="",
version="1.0.0",
)
# Get best run
best_run = utils.get_best_run(project="mahjouri-saamahn/mwml-httynn-app_v2",
metric="test_loss", objective="minimize")
# Load best run (if needed)
best_run_dir = utils.load_run(run=best_run)
# Get run components for prediction
args, model, X_tokenizer, y_tokenizer = predict.get_run_components(
run_dir=best_run_dir)
@utils.construct_response
@app.get("/")
async def _index():
response = {
'message': HTTPStatus.OK.phrase,
def normalize(x):
return (x - min(x)) / (max(x) - min(x))
# Title
st.title("Creating an End-to-End ML Application")
st.write(
"""[<img src="https://github.com/madewithml/images/blob/master/images/yt.png?raw=true" style="width:1.2rem;"> Watch Lesson](https://www.youtube.com/madewithml?sub_confirmation=1) ┬╖ [<img src="https://github.com/madewithml/images/blob/master/images/github_logo.png?raw=true" style="width:1.1rem;"> GitHub](https://github.com/madewithml/e2e-ml-app-tensorflow) ┬╖ [<img src="https://avatars0.githubusercontent.com/u/60439358?s=200&v=4" style="width:1.2rem;"> Made With ML](https://madewithml.com)""",
unsafe_allow_html=True)
st.write("Video lesson coming soon...")
# Get best run
project = 'GokuMohandas/e2e-ml-app-tensorflow'
best_run = utils.get_best_run(project=project,
metric="test_loss",
objective="minimize")
# Load best run (if needed)
best_run_dir = utils.load_run(run=best_run)
# Get run components for prediction
args, model, conv_outputs_model, X_tokenizer, y_tokenizer = predict.get_run_components(
run_dir=best_run_dir)
# Pages
page = st.sidebar.selectbox("Choose a page", ['Prediction', 'Model details'])
if page == 'Prediction':
st.header("ЁЯЪА Try it out!")
if __name__ == '__main__':
# Arguments
parser = ArgumentParser()
parser.add_argument('--text',
type=str,
required=True,
help="text to predict")
args = parser.parse_args()
inputs = [{'text': args.text}]
# Get best run
best_run = utils.get_best_run(
project="mahjouri-saamahn/mwml-app-tensorflow",
metric="test_loss",
objective="minimize")
# Load best run (if needed)
best_run_dir = utils.load_run(run=best_run)
# Get run components for prediction
args, model, conv_outputs_model, X_tokenizer, y_tokenizer = get_run_components(
run_dir=best_run_dir)
# Predict
results = predict(inputs=inputs,
args=args,
model=model,
conv_outputs_model=conv_outputs_model,
X_tokenizer=X_tokenizer,
conv_outputs=conv_outputs,
filter_sizes=args.filter_sizes)})
return results
if __name__ == '__main__':
# Arguments
parser = ArgumentParser()
parser.add_argument('--text', type=str,
required=True, help="text to predict")
args = parser.parse_args()
inputs = [{'text': args.text}]
# Get best run
best_run = utils.get_best_run(project="GokuMohandas/e2e-ml-app-tensorflow",
metric="test_loss", objective="minimize")
# Load best run (if needed)
best_run_dir = utils.load_run(run=best_run)
# Get run components for prediction
args, model, conv_outputs_model, X_tokenizer, y_tokenizer = get_run_components(
run_dir=best_run_dir)
# Predict
results = predict(inputs=inputs, args=args, model=model,
conv_outputs_model=conv_outputs_model,
X_tokenizer=X_tokenizer, y_tokenizer=y_tokenizer)
config.logger.info(json.dumps(results, indent=4, sort_keys=False))
results.append(performance)
return results
if __name__ == '__main__':
# Arguments
parser = ArgumentParser()
parser.add_argument('--text',
type=str,
required=True,
help="text to predict")
args = parser.parse_args()
inputs = [{'text': args.text}]
# Get best run
best_run = utils.get_best_run(project="mahjouri-saamahn/mwml-tutorial-app",
metric="test_loss",
objective="minimize")
# Load best run (if needed)
best_run_dir = utils.load_run(run=best_run)
# Get run components for prediction
model, word_map = get_run_components(run_dir=best_run_dir)
# Predict
results = predict(inputs=inputs, args=args, model=model, word_map=word_map)
config.logger.info(json.dumps(results, indent=4, sort_keys=False))
