Skip to content

GitLab

Unverified Commit 0c17e2d9 authored by chicm-ms's avatar chicm-ms Committed by GitHub
Browse files

Sdk update (#272) 

* Rename get_parameters to get_next_parameter

* annotations add get_next_parameter

* updates

* updates

* updates

* updates

* updates
parent a1014619
Hide whitespace changes
Inline Side-by-side
Showing with 327 additions and 314 deletions
docs/AnnotationSpec.md
View file @ 0c17e2d9
...@@ -4,23 +4,26 @@ For good user experience and reduce user effort, we need to design a good annota ...@@ -4,23 +4,26 @@ For good user experience and reduce user effort, we need to design a good annota
If users use NNI system, they only need to: If users use NNI system, they only need to:
1. Annotation variable in code as: 1. Use nni.get_next_parameter() to retrieve hyper parameters from Tuner, before using other annotation, use following annotation at the begining of trial code:
'''@nni.get_next_parameter()'''
2. Annotation variable in code as:
'''@nni.variable(nni.choice(2,3,5,7),name=self.conv_size)''' '''@nni.variable(nni.choice(2,3,5,7),name=self.conv_size)'''
2. Annotation intermediate in code as: 3. Annotation intermediate in code as:
'''@nni.report_intermediate_result(test_acc)''' '''@nni.report_intermediate_result(test_acc)'''
3. Annotation output in code as: 4. Annotation output in code as:
'''@nni.report_final_result(test_acc)''' '''@nni.report_final_result(test_acc)'''
4. Annotation `function_choice` in code as: 5. Annotation `function_choice` in code as:
'''@nni.function_choice(max_pool(h_conv1, self.pool_size),avg_pool(h_conv1, self.pool_size),name=max_pool)''' '''@nni.function_choice(max_pool(h_conv1, self.pool_size),avg_pool(h_conv1, self.pool_size),name=max_pool)'''
In this way, they can easily realize automatic tuning on NNI. In this way, they can easily implement automatic tuning on NNI.
For `@nni.variable`, `nni.choice` is the type of search space and there are 10 types to express your search space as follows: For `@nni.variable`, `nni.choice` is the type of search space and there are 10 types to express your search space as follows:
......
docs/howto_1_WriteTrial.md
View file @ 0c17e2d9
...@@ -27,7 +27,7 @@ Refer to [SearchSpaceSpec.md](SearchSpaceSpec.md) to learn more about search spa ...@@ -27,7 +27,7 @@ Refer to [SearchSpaceSpec.md](SearchSpaceSpec.md) to learn more about search spa
2.2 Get predefined parameters 2.2 Get predefined parameters
Use the following code snippet: Use the following code snippet:
RECEIVED_PARAMS = nni.get_parameters() RECEIVED_PARAMS = nni.get_next_parameter()
to get hyper-parameters' values assigned by tuner. `RECEIVED_PARAMS` is an object, for example: to get hyper-parameters' values assigned by tuner. `RECEIVED_PARAMS` is an object, for example:
......
docs/howto_2_CustomizedTuner.md
View file @ 0c17e2d9
...@@ -61,7 +61,7 @@ If the you implement the ```generate_parameters``` like this: ...@@ -61,7 +61,7 @@ If the you implement the ```generate_parameters``` like this:
# your code implements here. # your code implements here.
return {"dropout": 0.3, "learning_rate": 0.4} return {"dropout": 0.3, "learning_rate": 0.4}
``` ```
It's means your Tuner will always generate parameters ```{"dropout": 0.3, "learning_rate": 0.4}```. Then Trial will receive ```{"dropout": 0.3, "learning_rate": 0.4}``` this object will using ```nni.get_parameters()``` API from NNI SDK. After training of Trial, it will send result to Tuner by calling ```nni.report_final_result(0.93)```. Then ```receive_trial_result``` will function will receied these parameters like： It means your Tuner will always generate parameters ```{"dropout": 0.3, "learning_rate": 0.4}```. Then Trial will receive ```{"dropout": 0.3, "learning_rate": 0.4}``` by calling API ```nni.get_next_parameter()```. Once the trial ends with a result (normally some kind of metrics), it can send the result to Tuner by calling API ```nni.report_final_result()```, for example ```nni.report_final_result(0.93)```. Then your Tuner's ```receive_trial_result``` function will receied the result like：
``` ```
parameter_id = 82347 parameter_id = 82347
parameters = {"dropout": 0.3, "learning_rate": 0.4} parameters = {"dropout": 0.3, "learning_rate": 0.4}
......
examples/trials/README.md
View file @ 0c17e2d9
# How to write a Trial running on NNI? # How to write a Trial running on NNI?
*Trial receive the hyper-parameter/architecture configure from Tuner, and send intermediate result to Assessor and final result to Tuner.* *Trial receive the hyper-parameter/architecture configure from Tuner, and send intermediate result to Assessor and final result to Tuner.*
So when user want to write a Trial running on NNI, she/he should: So when user want to write a Trial running on NNI, she/he should:
**1)Have an original Trial could run**, **1)Have an original Trial could run**,
Trial's code could be any machine learning code that could run in local. Here we use ```mnist-keras.py``` as example: Trial's code could be any machine learning code that could run in local. Here we use ```mnist-keras.py``` as example:
```python ```python
import argparse import argparse
import logging import logging
import keras import keras
import numpy as np import numpy as np
from keras import backend as K from keras import backend as K
from keras.datasets import mnist from keras.datasets import mnist
from keras.layers import Conv2D, Dense, Flatten, MaxPooling2D from keras.layers import Conv2D, Dense, Flatten, MaxPooling2D
from keras.models import Sequential from keras.models import Sequential
K.set_image_data_format('channels_last') K.set_image_data_format('channels_last')
H, W = 28, 28 H, W = 28, 28
NUM_CLASSES = 10 NUM_CLASSES = 10
def create_mnist_model(hyper_params, input_shape=(H, W, 1), num_classes=NUM_CLASSES): def create_mnist_model(hyper_params, input_shape=(H, W, 1), num_classes=NUM_CLASSES):
layers = [ layers = [
Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=input_shape), Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=input_shape),
Conv2D(64, (3, 3), activation='relu'), Conv2D(64, (3, 3), activation='relu'),
MaxPooling2D(pool_size=(2, 2)), MaxPooling2D(pool_size=(2, 2)),
Flatten(), Flatten(),
Dense(100, activation='relu'), Dense(100, activation='relu'),
Dense(num_classes, activation='softmax') Dense(num_classes, activation='softmax')
] ]
model = Sequential(layers) model = Sequential(layers)
if hyper_params['optimizer'] == 'Adam': if hyper_params['optimizer'] == 'Adam':
optimizer = keras.optimizers.Adam(lr=hyper_params['learning_rate']) optimizer = keras.optimizers.Adam(lr=hyper_params['learning_rate'])
else: else:
optimizer = keras.optimizers.SGD(lr=hyper_params['learning_rate'], momentum=0.9) optimizer = keras.optimizers.SGD(lr=hyper_params['learning_rate'], momentum=0.9)
model.compile(loss=keras.losses.categorical_crossentropy, optimizer=optimizer, metrics=['accuracy']) model.compile(loss=keras.losses.categorical_crossentropy, optimizer=optimizer, metrics=['accuracy'])
return model return model
def load_mnist_data(args): def load_mnist_data(args):
(x_train, y_train), (x_test, y_test) = mnist.load_data() (x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = (np.expand_dims(x_train, -1).astype(np.float) / 255.)[:args.num_train] x_train = (np.expand_dims(x_train, -1).astype(np.float) / 255.)[:args.num_train]
x_test = (np.expand_dims(x_test, -1).astype(np.float) / 255.)[:args.num_test] x_test = (np.expand_dims(x_test, -1).astype(np.float) / 255.)[:args.num_test]
y_train = keras.utils.to_categorical(y_train, NUM_CLASSES)[:args.num_train] y_train = keras.utils.to_categorical(y_train, NUM_CLASSES)[:args.num_train]
y_test = keras.utils.to_categorical(y_test, NUM_CLASSES)[:args.num_test] y_test = keras.utils.to_categorical(y_test, NUM_CLASSES)[:args.num_test]
return x_train, y_train, x_test, y_test return x_train, y_train, x_test, y_test
class SendMetrics(keras.callbacks.Callback): class SendMetrics(keras.callbacks.Callback):
def on_epoch_end(self, epoch, logs={}): def on_epoch_end(self, epoch, logs={}):
pass pass
def train(args, params): def train(args, params):
x_train, y_train, x_test, y_test = load_mnist_data(args) x_train, y_train, x_test, y_test = load_mnist_data(args)
model = create_mnist_model(params) model = create_mnist_model(params)
model.fit(x_train, y_train, batch_size=args.batch_size, epochs=args.epochs, verbose=1, model.fit(x_train, y_train, batch_size=args.batch_size, epochs=args.epochs, verbose=1,
validation_data=(x_test, y_test), callbacks=[SendMetrics()]) validation_data=(x_test, y_test), callbacks=[SendMetrics()])
_, acc = model.evaluate(x_test, y_test, verbose=0) _, acc = model.evaluate(x_test, y_test, verbose=0)
def generate_default_params(): def generate_default_params():
return { return {
'optimizer': 'Adam', 'optimizer': 'Adam',
'learning_rate': 0.001 'learning_rate': 0.001
} }
if __name__ == '__main__': if __name__ == '__main__':
PARSER = argparse.ArgumentParser() PARSER = argparse.ArgumentParser()
PARSER.add_argument("--batch_size", type=int, default=200, help="batch size", required=False) PARSER.add_argument("--batch_size", type=int, default=200, help="batch size", required=False)
PARSER.add_argument("--epochs", type=int, default=10, help="Train epochs", required=False) PARSER.add_argument("--epochs", type=int, default=10, help="Train epochs", required=False)
PARSER.add_argument("--num_train", type=int, default=1000, help="Number of train samples to be used, maximum 60000", required=False) PARSER.add_argument("--num_train", type=int, default=1000, help="Number of train samples to be used, maximum 60000", required=False)
PARSER.add_argument("--num_test", type=int, default=1000, help="Number of test samples to be used, maximum 10000", required=False) PARSER.add_argument("--num_test", type=int, default=1000, help="Number of test samples to be used, maximum 10000", required=False)
ARGS, UNKNOWN = PARSER.parse_known_args() ARGS, UNKNOWN = PARSER.parse_known_args()
PARAMS = generate_default_params() PARAMS = generate_default_params()
train(ARGS, PARAMS) train(ARGS, PARAMS)
``` ```
**2)Get configure from Tuner** **2)Get configure from Tuner**
User import ```nni``` and use ```nni.get_parameters()``` to recive configure. Please noted **10**, **24** and **25** line in the following code. User import ```nni``` and use ```nni.get_next_parameter()``` to recive configure. Please noted **10**, **24** and **25** line in the following code.
```python ```python
import argparse import argparse
import logging import logging
import keras import keras
import numpy as np import numpy as np
from keras import backend as K from keras import backend as K
from keras.datasets import mnist from keras.datasets import mnist
from keras.layers import Conv2D, Dense, Flatten, MaxPooling2D from keras.layers import Conv2D, Dense, Flatten, MaxPooling2D
from keras.models import Sequential from keras.models import Sequential
import nni import nni
... ...
if __name__ == '__main__': if __name__ == '__main__':
PARSER = argparse.ArgumentParser() PARSER = argparse.ArgumentParser()
PARSER.add_argument("--batch_size", type=int, default=200, help="batch size", required=False) PARSER.add_argument("--batch_size", type=int, default=200, help="batch size", required=False)
PARSER.add_argument("--epochs", type=int, default=10, help="Train epochs", required=False) PARSER.add_argument("--epochs", type=int, default=10, help="Train epochs", required=False)
PARSER.add_argument("--num_train", type=int, default=1000, help="Number of train samples to be used, maximum 60000", required=False) PARSER.add_argument("--num_train", type=int, default=1000, help="Number of train samples to be used, maximum 60000", required=False)
PARSER.add_argument("--num_test", type=int, default=1000, help="Number of test samples to be used, maximum 10000", required=False) PARSER.add_argument("--num_test", type=int, default=1000, help="Number of test samples to be used, maximum 10000", required=False)
ARGS, UNKNOWN = PARSER.parse_known_args() ARGS, UNKNOWN = PARSER.parse_known_args()
PARAMS = generate_default_params() PARAMS = generate_default_params()
RECEIVED_PARAMS = nni.get_parameters() RECEIVED_PARAMS = nni.get_next_parameter()
PARAMS.update(RECEIVED_PARAMS) PARAMS.update(RECEIVED_PARAMS)
train(ARGS, PARAMS) train(ARGS, PARAMS)
``` ```
**3) Send intermediate result** **3) Send intermediate result**
Use ```nni.report_intermediate_result``` to send intermediate result to Assessor. Please noted **5** line in the following code. Use ```nni.report_intermediate_result``` to send intermediate result to Assessor. Please noted **5** line in the following code.
```python ```python
... ...
class SendMetrics(keras.callbacks.Callback): class SendMetrics(keras.callbacks.Callback):
def on_epoch_end(self, epoch, logs={}): def on_epoch_end(self, epoch, logs={}):
nni.report_intermediate_result(logs) nni.report_intermediate_result(logs)
def train(args, params): def train(args, params):
x_train, y_train, x_test, y_test = load_mnist_data(args) x_train, y_train, x_test, y_test = load_mnist_data(args)
model = create_mnist_model(params) model = create_mnist_model(params)
model.fit(x_train, y_train, batch_size=args.batch_size, epochs=args.epochs, verbose=1, model.fit(x_train, y_train, batch_size=args.batch_size, epochs=args.epochs, verbose=1,
validation_data=(x_test, y_test), callbacks=[SendMetrics()]) validation_data=(x_test, y_test), callbacks=[SendMetrics()])
_, acc = model.evaluate(x_test, y_test, verbose=0) _, acc = model.evaluate(x_test, y_test, verbose=0)
... ...
``` ```
**4) Send final result** **4) Send final result**
Use ```nni.report_final_result``` to send final result to Trial. Please noted **15** line in the following code. Use ```nni.report_final_result``` to send final result to Trial. Please noted **15** line in the following code.
```python ```python
... ...
class SendMetrics(keras.callbacks.Callback): class SendMetrics(keras.callbacks.Callback):
def on_epoch_end(self, epoch, logs={}): def on_epoch_end(self, epoch, logs={}):
nni.report_intermediate_result(logs) nni.report_intermediate_result(logs)
def train(args, params): def train(args, params):
x_train, y_train, x_test, y_test = load_mnist_data(args) x_train, y_train, x_test, y_test = load_mnist_data(args)
model = create_mnist_model(params) model = create_mnist_model(params)
model.fit(x_train, y_train, batch_size=args.batch_size, epochs=args.epochs, verbose=1, model.fit(x_train, y_train, batch_size=args.batch_size, epochs=args.epochs, verbose=1,
validation_data=(x_test, y_test), callbacks=[SendMetrics()]) validation_data=(x_test, y_test), callbacks=[SendMetrics()])
_, acc = model.evaluate(x_test, y_test, verbose=0) _, acc = model.evaluate(x_test, y_test, verbose=0)
nni.report_final_result(acc) nni.report_final_result(acc)
... ...
``` ```
Here is the complete exampe: Here is the complete exampe:
```python ```python
import argparse import argparse
import logging import logging
import keras import keras
import numpy as np import numpy as np
from keras import backend as K from keras import backend as K
from keras.datasets import mnist from keras.datasets import mnist
from keras.layers import Conv2D, Dense, Flatten, MaxPooling2D from keras.layers import Conv2D, Dense, Flatten, MaxPooling2D
from keras.models import Sequential from keras.models import Sequential
import nni import nni
LOG = logging.getLogger('mnist_keras') LOG = logging.getLogger('mnist_keras')
K.set_image_data_format('channels_last') K.set_image_data_format('channels_last')
H, W = 28, 28 H, W = 28, 28
NUM_CLASSES = 10 NUM_CLASSES = 10
def create_mnist_model(hyper_params, input_shape=(H, W, 1), num_classes=NUM_CLASSES): def create_mnist_model(hyper_params, input_shape=(H, W, 1), num_classes=NUM_CLASSES):
''' '''
Create simple convolutional model Create simple convolutional model
''' '''
layers = [ layers = [
Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=input_shape), Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=input_shape),
Conv2D(64, (3, 3), activation='relu'), Conv2D(64, (3, 3), activation='relu'),
MaxPooling2D(pool_size=(2, 2)), MaxPooling2D(pool_size=(2, 2)),
Flatten(), Flatten(),
Dense(100, activation='relu'), Dense(100, activation='relu'),
Dense(num_classes, activation='softmax') Dense(num_classes, activation='softmax')
] ]
model = Sequential(layers) model = Sequential(layers)
if hyper_params['optimizer'] == 'Adam': if hyper_params['optimizer'] == 'Adam':
optimizer = keras.optimizers.Adam(lr=hyper_params['learning_rate']) optimizer = keras.optimizers.Adam(lr=hyper_params['learning_rate'])
else: else:
optimizer = keras.optimizers.SGD(lr=hyper_params['learning_rate'], momentum=0.9) optimizer = keras.optimizers.SGD(lr=hyper_params['learning_rate'], momentum=0.9)
model.compile(loss=keras.losses.categorical_crossentropy, optimizer=optimizer, metrics=['accuracy']) model.compile(loss=keras.losses.categorical_crossentropy, optimizer=optimizer, metrics=['accuracy'])
return model return model
def load_mnist_data(args): def load_mnist_data(args):
''' '''
Load MNIST dataset Load MNIST dataset
''' '''
(x_train, y_train), (x_test, y_test) = mnist.load_data() (x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = (np.expand_dims(x_train, -1).astype(np.float) / 255.)[:args.num_train] x_train = (np.expand_dims(x_train, -1).astype(np.float) / 255.)[:args.num_train]
x_test = (np.expand_dims(x_test, -1).astype(np.float) / 255.)[:args.num_test] x_test = (np.expand_dims(x_test, -1).astype(np.float) / 255.)[:args.num_test]
y_train = keras.utils.to_categorical(y_train, NUM_CLASSES)[:args.num_train] y_train = keras.utils.to_categorical(y_train, NUM_CLASSES)[:args.num_train]
y_test = keras.utils.to_categorical(y_test, NUM_CLASSES)[:args.num_test] y_test = keras.utils.to_categorical(y_test, NUM_CLASSES)[:args.num_test]
LOG.debug('x_train shape: %s', (x_train.shape,)) LOG.debug('x_train shape: %s', (x_train.shape,))
LOG.debug('x_test shape: %s', (x_test.shape,)) LOG.debug('x_test shape: %s', (x_test.shape,))
return x_train, y_train, x_test, y_test return x_train, y_train, x_test, y_test
class SendMetrics(keras.callbacks.Callback): class SendMetrics(keras.callbacks.Callback):
''' '''
Keras callback to send metrics to NNI framework Keras callback to send metrics to NNI framework
''' '''
def on_epoch_end(self, epoch, logs={}): def on_epoch_end(self, epoch, logs={}):
''' '''
Run on end of each epoch Run on end of each epoch
''' '''
LOG.debug(logs) LOG.debug(logs)
nni.report_intermediate_result(logs) nni.report_intermediate_result(logs)
def train(args, params): def train(args, params):
''' '''
Train model Train model
''' '''
x_train, y_train, x_test, y_test = load_mnist_data(args) x_train, y_train, x_test, y_test = load_mnist_data(args)
model = create_mnist_model(params) model = create_mnist_model(params)
model.fit(x_train, y_train, batch_size=args.batch_size, epochs=args.epochs, verbose=1, model.fit(x_train, y_train, batch_size=args.batch_size, epochs=args.epochs, verbose=1,
validation_data=(x_test, y_test), callbacks=[SendMetrics()]) validation_data=(x_test, y_test), callbacks=[SendMetrics()])
_, acc = model.evaluate(x_test, y_test, verbose=0) _, acc = model.evaluate(x_test, y_test, verbose=0)
LOG.debug('Final result is: %d', acc) LOG.debug('Final result is: %d', acc)
nni.report_final_result(acc) nni.report_final_result(acc)
def generate_default_params(): def generate_default_params():
''' '''
Generate default hyper parameters Generate default hyper parameters
''' '''
return { return {
'optimizer': 'Adam', 'optimizer': 'Adam',
'learning_rate': 0.001 'learning_rate': 0.001
} }
if __name__ == '__main__': if __name__ == '__main__':
PARSER = argparse.ArgumentParser() PARSER = argparse.ArgumentParser()
PARSER.add_argument("--batch_size", type=int, default=200, help="batch size", required=False) PARSER.add_argument("--batch_size", type=int, default=200, help="batch size", required=False)
PARSER.add_argument("--epochs", type=int, default=10, help="Train epochs", required=False) PARSER.add_argument("--epochs", type=int, default=10, help="Train epochs", required=False)
PARSER.add_argument("--num_train", type=int, default=1000, help="Number of train samples to be used, maximum 60000", required=False) PARSER.add_argument("--num_train", type=int, default=1000, help="Number of train samples to be used, maximum 60000", required=False)
PARSER.add_argument("--num_test", type=int, default=1000, help="Number of test samples to be used, maximum 10000", required=False) PARSER.add_argument("--num_test", type=int, default=1000, help="Number of test samples to be used, maximum 10000", required=False)
ARGS, UNKNOWN = PARSER.parse_known_args() ARGS, UNKNOWN = PARSER.parse_known_args()
try: try:
# get parameters from tuner # get parameters from tuner
RECEIVED_PARAMS = nni.get_parameters() RECEIVED_PARAMS = nni.get_next_parameter()
LOG.debug(RECEIVED_PARAMS) LOG.debug(RECEIVED_PARAMS)
PARAMS = generate_default_params() PARAMS = generate_default_params()
PARAMS.update(RECEIVED_PARAMS) PARAMS.update(RECEIVED_PARAMS)
# train # train
train(ARGS, PARAMS) train(ARGS, PARAMS)
except Exception as e: except Exception as e:
LOG.exception(e) LOG.exception(e)
raise raise
``` ```
\ No newline at end of file
examples/trials/auto-gbdt/main.py
View file @ 0c17e2d9
...@@ -97,7 +97,7 @@ if __name__ == '__main__': ...@@ -97,7 +97,7 @@ if __name__ == '__main__':
try: try:
# get parameters from tuner # get parameters from tuner
RECEIVED_PARAMS = nni.get_parameters() RECEIVED_PARAMS = nni.get_next_parameter()
LOG.debug(RECEIVED_PARAMS) LOG.debug(RECEIVED_PARAMS)
PARAMS = get_default_parameters() PARAMS = get_default_parameters()
PARAMS.update(RECEIVED_PARAMS) PARAMS.update(RECEIVED_PARAMS)
......
examples/trials/ga_squad/trial.py
View file @ 0c17e2d9
...@@ -436,7 +436,7 @@ if __name__ == '__main__': ...@@ -436,7 +436,7 @@ if __name__ == '__main__':
qp_pairs, dev_qp_pairs = load_data() qp_pairs, dev_qp_pairs = load_data()
logger.debug('Init finish.') logger.debug('Init finish.')
original_params = nni.get_parameters() original_params = nni.get_next_parameter()
''' '''
with open('data.json') as f: with open('data.json') as f:
original_params = json.load(f) original_params = json.load(f)
......
examples/trials/mnist-annotation/mnist.py
View file @ 0c17e2d9
...@@ -229,6 +229,7 @@ def generate_defualt_params(): ...@@ -229,6 +229,7 @@ def generate_defualt_params():
if __name__ == '__main__': if __name__ == '__main__':
'''@nni.get_next_parameter()'''
try: try:
main(generate_defualt_params()) main(generate_defualt_params())
except Exception as exception: except Exception as exception:
......
examples/trials/mnist-batch-tune-keras/mnist-keras.py
View file @ 0c17e2d9
...@@ -122,7 +122,7 @@ if __name__ == '__main__': ...@@ -122,7 +122,7 @@ if __name__ == '__main__':
try: try:
# get parameters from tuner # get parameters from tuner
# RECEIVED_PARAMS = {"optimizer": "Adam", "learning_rate": 0.00001} # RECEIVED_PARAMS = {"optimizer": "Adam", "learning_rate": 0.00001}
RECEIVED_PARAMS = nni.get_parameters() RECEIVED_PARAMS = nni.get_next_parameter()
LOG.debug(RECEIVED_PARAMS) LOG.debug(RECEIVED_PARAMS)
PARAMS = generate_default_params() PARAMS = generate_default_params()
PARAMS.update(RECEIVED_PARAMS) PARAMS.update(RECEIVED_PARAMS)
......
examples/trials/mnist-cascading-search-space/mnist.py
View file @ 0c17e2d9
...@@ -149,7 +149,7 @@ def parse_init_json(data): ...@@ -149,7 +149,7 @@ def parse_init_json(data):
if __name__ == '__main__': if __name__ == '__main__':
try: try:
# get parameters form tuner # get parameters form tuner
data = nni.get_parameters() data = nni.get_next_parameter()
logger.debug(data) logger.debug(data)
RCV_PARAMS = parse_init_json(data) RCV_PARAMS = parse_init_json(data)
......
examples/trials/mnist-keras/mnist-keras.py
View file @ 0c17e2d9
...@@ -120,7 +120,7 @@ if __name__ == '__main__': ...@@ -120,7 +120,7 @@ if __name__ == '__main__':
try: try:
# get parameters from tuner # get parameters from tuner
RECEIVED_PARAMS = nni.get_parameters() RECEIVED_PARAMS = nni.get_next_parameter()
LOG.debug(RECEIVED_PARAMS) LOG.debug(RECEIVED_PARAMS)
PARAMS = generate_default_params() PARAMS = generate_default_params()
PARAMS.update(RECEIVED_PARAMS) PARAMS.update(RECEIVED_PARAMS)
......
examples/trials/mnist/mnist.py
View file @ 0c17e2d9
...@@ -219,7 +219,7 @@ def generate_default_params(): ...@@ -219,7 +219,7 @@ def generate_default_params():
if __name__ == '__main__': if __name__ == '__main__':
try: try:
# get parameters form tuner # get parameters form tuner
RCV_PARAMS = nni.get_parameters() RCV_PARAMS = nni.get_next_parameter()
logger.debug(RCV_PARAMS) logger.debug(RCV_PARAMS)
# run # run
params = generate_default_params() params = generate_default_params()
......
examples/trials/pytorch_cifar10/main.py
View file @ 0c17e2d9
...@@ -175,7 +175,7 @@ def test(epoch): ...@@ -175,7 +175,7 @@ def test(epoch):
if __name__ == '__main__': if __name__ == '__main__':
try: try:
RCV_CONFIG = nni.get_parameters() RCV_CONFIG = nni.get_next_parameter()
#RCV_CONFIG = {'lr': 0.1, 'optimizer': 'Adam', 'model':'senet18'} #RCV_CONFIG = {'lr': 0.1, 'optimizer': 'Adam', 'model':'senet18'}
_logger.debug(RCV_CONFIG) _logger.debug(RCV_CONFIG)
......
examples/trials/sklearn/classification/main.py
View file @ 0c17e2d9
...@@ -71,7 +71,7 @@ if __name__ == '__main__': ...@@ -71,7 +71,7 @@ if __name__ == '__main__':
try: try:
# get parameters from tuner # get parameters from tuner
RECEIVED_PARAMS = nni.get_parameters() RECEIVED_PARAMS = nni.get_next_parameter()
LOG.debug(RECEIVED_PARAMS) LOG.debug(RECEIVED_PARAMS)
PARAMS = get_default_parameters() PARAMS = get_default_parameters()
PARAMS.update(RECEIVED_PARAMS) PARAMS.update(RECEIVED_PARAMS)
......
examples/trials/sklearn/regression/main.py
View file @ 0c17e2d9
...@@ -90,7 +90,7 @@ if __name__ == '__main__': ...@@ -90,7 +90,7 @@ if __name__ == '__main__':
try: try:
# get parameters from tuner # get parameters from tuner
RECEIVED_PARAMS = nni.get_parameters() RECEIVED_PARAMS = nni.get_next_parameter()
LOG.debug(RECEIVED_PARAMS) LOG.debug(RECEIVED_PARAMS)
PARAMS = get_default_parameters() PARAMS = get_default_parameters()
PARAMS.update(RECEIVED_PARAMS) PARAMS.update(RECEIVED_PARAMS)
......
src/sdk/pynni/nni/platform/local.py
View file @ 0c17e2d9
...@@ -49,13 +49,18 @@ def request_next_parameter(): ...@@ -49,13 +49,18 @@ def request_next_parameter():
}) })
send_metric(metric) send_metric(metric)
def get_parameters(): def get_next_parameter():
global _param_index global _param_index
params_file_name = '' params_file_name = ''
if _multiphase and (_multiphase == 'true' or _multiphase == 'True'): if _multiphase and (_multiphase == 'true' or _multiphase == 'True'):
params_file_name = ('parameter_{}.cfg'.format(_param_index), 'parameter.cfg')[_param_index == 0] params_file_name = ('parameter_{}.cfg'.format(_param_index), 'parameter.cfg')[_param_index == 0]
else: else:
params_file_name = 'parameter.cfg' if _param_index > 0:
return None
elif _param_index == 0:
params_file_name = 'parameter.cfg'
else:
raise AssertionError('_param_index value ({}) should >=0'.format(_param_index))
params_filepath = os.path.join(_sysdir, params_file_name) params_filepath = os.path.join(_sysdir, params_file_name)
if not os.path.isfile(params_filepath): if not os.path.isfile(params_filepath):
......
src/sdk/pynni/nni/platform/standalone.py
View file @ 0c17e2d9
...@@ -22,7 +22,7 @@ ...@@ -22,7 +22,7 @@
import json_tricks import json_tricks
def get_parameters(): def get_next_parameter():
pass pass
def get_sequence_id(): def get_sequence_id():
......
src/sdk/pynni/nni/platform/test.py
View file @ 0c17e2d9
...@@ -29,7 +29,7 @@ _params = None ...@@ -29,7 +29,7 @@ _params = None
_last_metric = None _last_metric = None
def get_parameters(): def get_next_parameter():
return _params return _params
def send_metric(string): def send_metric(string):
......
src/sdk/pynni/nni/smartparam.py
View file @ 0c17e2d9
...@@ -126,4 +126,4 @@ else: ...@@ -126,4 +126,4 @@ else:
if name is None: if name is None:
name = '__line{:d}'.format(lineno) name = '__line{:d}'.format(lineno)
key = '{}/{}/{}'.format(module, name, func) key = '{}/{}/{}'.format(module, name, func)
return trial.get_parameter(key) return trial.get_current_parameter(key)
src/sdk/pynni/nni/trial.py
View file @ 0c17e2d9
...@@ -26,7 +26,8 @@ from . import platform ...@@ -26,7 +26,8 @@ from . import platform
__all__ = [ __all__ = [
'get_parameters', 'get_next_parameter',
'get_current_parameter',
'report_intermediate_result', 'report_intermediate_result',
'report_final_result', 'report_final_result',
'get_sequence_id' 'get_sequence_id'
...@@ -37,15 +38,18 @@ _params = None ...@@ -37,15 +38,18 @@ _params = None
_sequence_id = platform.get_sequence_id() _sequence_id = platform.get_sequence_id()
def get_parameters(): def get_next_parameter():
"""Returns a set of (hyper-)paremeters generated by Tuner.""" """Returns a set of (hyper-)paremeters generated by Tuner."""
global _params global _params
_params = platform.get_parameters() _params = platform.get_next_parameter()
if _params is None:
return None
return _params['parameters'] return _params['parameters']
def get_current_parameter(tag):
def get_parameter(tag): if _params is None:
return get_parameters()[tag] return None
return _params['parameters'][tag]
def get_sequence_id(): def get_sequence_id():
return _sequence_id return _sequence_id
...@@ -57,7 +61,7 @@ def report_intermediate_result(metric): ...@@ -57,7 +61,7 @@ def report_intermediate_result(metric):
metric: serializable object. metric: serializable object.
""" """
global _intermediate_seq global _intermediate_seq
assert _params is not None, 'nni.get_parameters() needs to be called before report_intermediate_result' assert _params is not None, 'nni.get_next_parameter() needs to be called before report_intermediate_result'
metric = json_tricks.dumps({ metric = json_tricks.dumps({
'parameter_id': _params['parameter_id'], 'parameter_id': _params['parameter_id'],
'trial_job_id': env_args.trial_job_id, 'trial_job_id': env_args.trial_job_id,
...@@ -73,7 +77,7 @@ def report_final_result(metric): ...@@ -73,7 +77,7 @@ def report_final_result(metric):
"""Reports final result to tuner. """Reports final result to tuner.
metric: serializable object. metric: serializable object.
""" """
assert _params is not None, 'nni.get_parameters() needs to be called before report_final_result' assert _params is not None, 'nni.get_next_parameter() needs to be called before report_final_result'
metric = json_tricks.dumps({ metric = json_tricks.dumps({
'parameter_id': _params['parameter_id'], 'parameter_id': _params['parameter_id'],
'trial_job_id': env_args.trial_job_id, 'trial_job_id': env_args.trial_job_id,
......
src/sdk/pynni/tests/test_trial.py
View file @ 0c17e2d9
...@@ -32,8 +32,8 @@ class TrialTestCase(TestCase): ...@@ -32,8 +32,8 @@ class TrialTestCase(TestCase):
self._trial_params = { 'msg': 'hi', 'x': 123, 'dict': { 'key': 'value', 'y': None } } self._trial_params = { 'msg': 'hi', 'x': 123, 'dict': { 'key': 'value', 'y': None } }
nni.trial._params = { 'parameter_id': 'test_param', 'parameters': self._trial_params } nni.trial._params = { 'parameter_id': 'test_param', 'parameters': self._trial_params }
def test_get_parameters(self): def test_get_next_parameter(self):
self.assertEqual(nni.get_parameters(), self._trial_params) self.assertEqual(nni.get_next_parameter(), self._trial_params)
def test_report_intermediate_result(self): def test_report_intermediate_result(self):
nni.report_intermediate_result(123) nni.report_intermediate_result(123)
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment