def check(event, context):
try:
# Fetch data
data = fetch_post_data(event)
infer_config = fetch_inference_json()
# Check if model is currently training
server_status_config = fetch_status()
if server_status_config['status'] == 'active' and server_status_config['token'] == data['token']:
return create_response({
'result': 'error',
'message': 'The model is currently training. Please try again after a few minutes.'
})
# Check if token exists
if not data['token'] in infer_config:
return create_response({
'result': 'error',
'message': 'No such token found.'
})
task_config = infer_config[data['token']]
return create_response({
'result': 'success',
'taskType': task_config['task_type'],
'accuracy': task_config['accuracy'],
'accuracyPlot': task_config['accuracy_plot'],
})
except Exception as e:
print(repr(e))
return create_response({
'result': 'internal_error',
'message': repr(e),
}, status_code=500)
def train(event, context):
try:
server_status = fetch_status()
if server_status['status'] == 'active':
return create_response({
'result': 'error',
'message': 'Server is busy.',
})
# Fetch data
data = fetch_post_data(event)
# Check if server is properly shutdown
if EC2_RESOURCE.Instance(INSTANCE_ID).state['Name'] == 'stopping':
return create_response({
'result': 'error',
'message': 'Server is currently training another model, please check back in 5 minutes.'
})
# Get number of classes and validate data
if data['taskType'].lower() == 'sentimentanalysis':
validation_response = validate_csv(data['dataset'])
if validation_response['is_valid']:
data['dataset'] = validation_response['data']
data['numClasses'] = validation_response['num_classes']
else:
return create_response({
'result': 'error',
'message': validation_response['message'],
})
else:
data['numClasses'] = len(data['dataset'])
# Create token
token = create_user_token(
data['taskType'], data['taskName']
)
print('Token:', token)
# Change server status to active
change_server_status(
'active', server_status['dev_mode'],
task_type=data['taskType'].lower(), token=token
)
# Initialize training process
create_training_json(token, data)
return create_response({
'result': 'success',
'token': token
})
except Exception as e:
print(repr(e))
return create_response({
'result': 'internal_error',
'message': repr(e),
}, status_code=500)
def status(event, context):
try:
return create_response({
'result': 'success',
'status': fetch_status()['status'],
})
except Exception as e:
print(repr(e))
return create_response({
'result': 'internal_error',
'message': repr(e),
}, status_code=500)
def server_start(event, context):
message = 'Status not active. Server not turned on.'
server_status = fetch_status()
if server_status['dev_mode']:
message = 'Dev mode is on.'
elif server_status['status'] == 'active':
ec2_client = boto3.client('ec2', region_name=REGION)
ec2_client.start_instances(InstanceIds=[INSTANCE_ID])
message = 'Instance started.'
print(message)
return create_response({
'message': message
})
def server_stop(event, context):
server_status = fetch_status()
if server_status['dev_mode']:
message = 'Dev mode is on.'
else:
# Stop instance
ec2_client = boto3.client('ec2', region_name=REGION)
ec2_client.stop_instances(InstanceIds=[INSTANCE_ID])
message = 'Instance stopped.'
# Change server status
change_server_status('sleeping', server_status['dev_mode'])
print(message)
return create_response({
'message': message
})