iicd/xautodl
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

update NAS-Bench-102

Browse Source
This commit is contained in:
D-X-Y 2019-12-21 11:13:08 +11:00
parent 69ca0860aa
commit 95ec4d328e
3 changed files with 76 additions and 39 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
.gitignore vendored
View File

@ -115,3 +115,6 @@ GPU-*.sh
cal.sh cal.sh
aaa aaa
cx.sh cx.sh
NAS-Bench-102-v1_0.pth
lib/NAS-Bench-102-v1_0.pth

19
NAS-Bench-102.md
View File

@ -6,11 +6,16 @@ Each edge here is associated with an operation selected from a predefined operat
For it to be applicable for all NAS algorithms, the search space defined in NAS-Bench-102 includes 4 nodes and 5 associated operation options, which generates 15,625 neural cell candidates in total. For it to be applicable for all NAS algorithms, the search space defined in NAS-Bench-102 includes 4 nodes and 5 associated operation options, which generates 15,625 neural cell candidates in total.
In this Markdown file, we provide: In this Markdown file, we provide:
- Detailed instruction to reproduce NAS-Bench-102. - [How to Use NAS-Bench-102](#how-to-use-nas-bench-102)
- 10 NAS algorithms evaluated in our paper. - [Instruction to re-generate NAS-Bench-102](#instruction-to-re-generate-nas-bench-102)
- [10 NAS algorithms evaluated in our paper](#to-reproduce-10-baseline-nas-algorithms-in-nas-bench-102)
Note: please use `PyTorch >= 1.2.0` and `Python >= 3.6.0`. Note: please use `PyTorch >= 1.2.0` and `Python >= 3.6.0`.
The data file of NAS-Bench-102 can be downloaded from [Google Drive](https://drive.google.com/open?id=1SKW0Cu0u8-gb18zDpaAGi0f74UdXeGKs) or [Baidu-Wangpan].
The training and evaluation data used in NAS-Bench-102 can be downloaded from [Google Drive](https://drive.google.com/open?id=1L0Lzq8rWpZLPfiQGd6QR8q5xLV88emU7) or [Baidu-Wangpan].
## How to Use NAS-Bench-102 ## How to Use NAS-Bench-102
1. Creating an API instance from a file: 1. Creating an API instance from a file:
@ -35,8 +40,8 @@ api.show(2)
# show the mean loss and accuracy of an architecture # show the mean loss and accuracy of an architecture
info = api.query_meta_info_by_index(1) info = api.query_meta_info_by_index(1)
loss, accuracy = info.get_metrics('cifar10', 'train') res_metrics = info.get_metrics('cifar10', 'train')
flops, params, latency = info.get_comput_costs('cifar100') cost_metrics = info.get_comput_costs('cifar100')
# get the detailed information # get the detailed information
results = api.query_by_index(1, 'cifar100') results = api.query_by_index(1, 'cifar100')
@ -55,7 +60,8 @@ index = api.query_index_by_arch('|nor_conv_3x3~0|+|nor_conv_3x3~0|avg_pool_3x3~1
api.show(index) api.show(index)
``` ```
5. For other usages, please see `lib/aa_nas_api/api.py` 5. For other usages, please see `lib/nas_102_api/api.py`
### Detailed Instruction ### Detailed Instruction
@ -98,8 +104,10 @@ print(archRes.get_metrics('cifar10-valid', 'x-valid', None, True)) # print loss
``` ```
from nas_102_api import NASBench102API as API from nas_102_api import NASBench102API as API
api = API('NAS-Bench-102-v1_0.pth') api = API('NAS-Bench-102-v1_0.pth')
api.show(-1) # show info of all architectures
``` ```
## Instruction to Re-Generate NAS-Bench-102 ## Instruction to Re-Generate NAS-Bench-102
1. generate the meta file for NAS-Bench-102 using the following script, where `NAS-BENCH-102` indicates the name and `4` indicates the maximum number of nodes in a cell. 1. generate the meta file for NAS-Bench-102 using the following script, where `NAS-BENCH-102` indicates the name and `4` indicates the maximum number of nodes in a cell.
@ -139,6 +147,7 @@ CUDA_VISIBLE_DEVICES=0 bash ./scripts-search/NAS-Bench-102/train-a-net.sh resnet
CUDA_VISIBLE_DEVICES=0 bash ./scripts-search/NAS-Bench-102/train-a-net.sh '|nor_conv_3x3~0|+|nor_conv_3x3~0|nor_conv_3x3~1|+|skip_connect~0|skip_connect~1|skip_connect~2|' 16 5 CUDA_VISIBLE_DEVICES=0 bash ./scripts-search/NAS-Bench-102/train-a-net.sh '|nor_conv_3x3~0|+|nor_conv_3x3~0|nor_conv_3x3~1|+|skip_connect~0|skip_connect~1|skip_connect~2|' 16 5
``` ```
## To Reproduce 10 Baseline NAS Algorithms in NAS-Bench-102 ## To Reproduce 10 Baseline NAS Algorithms in NAS-Bench-102
We have tried our best to implement each method. However, still, some algorithms might obtain non-optimal results since their hyper-parameters might not fit our NAS-Bench-102. We have tried our best to implement each method. However, still, some algorithms might obtain non-optimal results since their hyper-parameters might not fit our NAS-Bench-102.

91
lib/nas_102_api/api.py
View File

@ -1,6 +1,8 @@
################################################## ##################################################
# Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019 # # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019 #
################################################## #################################################################################
# NAS-Bench-102: Extending the Scope of Reproducible Neural Architecture Search #
#################################################################################
import os, sys, copy, random, torch, numpy as np import os, sys, copy, random, torch, numpy as np
from collections import OrderedDict, defaultdict from collections import OrderedDict, defaultdict
@ -12,19 +14,21 @@ def print_information(information, extra_info=None, show=False):
return 'loss = {:.3f}, top1 = {:.2f}%'.format(loss, acc) return 'loss = {:.3f}, top1 = {:.2f}%'.format(loss, acc)
for ida, dataset in enumerate(dataset_names): for ida, dataset in enumerate(dataset_names):
flop, param, latency = information.get_comput_costs(dataset) #flop, param, latency = information.get_comput_costs(dataset)
str1 = '{:14s} FLOP={:6.2f} M, Params={:.3f} MB, latency={:} ms.'.format(dataset, flop, param, '{:.2f}'.format(latency*1000) if latency > 0 else None) metric = information.get_comput_costs(dataset)
train_loss, train_acc = information.get_metrics(dataset, 'train') flop, param, latency = metric['flops'], metric['params'], metric['latency']
str1 = '{:14s} FLOP={:6.2f} M, Params={:.3f} MB, latency={:} ms.'.format(dataset, flop, param, '{:.2f}'.format(latency*1000) if latency is not None and latency > 0 else None)
train_info = information.get_metrics(dataset, 'train')
if dataset == 'cifar10-valid': if dataset == 'cifar10-valid':
valid_loss, valid_acc = information.get_metrics(dataset, 'x-valid') valid_info = information.get_metrics(dataset, 'x-valid')
str2 = '{:14s} train : [{:}], valid : [{:}]'.format(dataset, metric2str(train_loss, train_acc), metric2str(valid_loss, valid_acc)) str2 = '{:14s} train : [{:}], valid : [{:}]'.format(dataset, metric2str(train_info['loss'], train_info['accuracy']), metric2str(valid_info['loss'], valid_info['accuracy']))
elif dataset == 'cifar10': elif dataset == 'cifar10':
test__loss, test__acc = information.get_metrics(dataset, 'ori-test') test__info = information.get_metrics(dataset, 'ori-test')
str2 = '{:14s} train : [{:}], test : [{:}]'.format(dataset, metric2str(train_loss, train_acc), metric2str(test__loss, test__acc)) str2 = '{:14s} train : [{:}], test : [{:}]'.format(dataset, metric2str(train_info['loss'], train_info['accuracy']), metric2str(test__info['loss'], test__info['accuracy']))
else: else:
valid_loss, valid_acc = information.get_metrics(dataset, 'x-valid') valid_info = information.get_metrics(dataset, 'x-valid')
test__loss, test__acc = information.get_metrics(dataset, 'x-test') test__info = information.get_metrics(dataset, 'x-test')
str2 = '{:14s} train : [{:}], valid : [{:}], test : [{:}]'.format(dataset, metric2str(train_loss, train_acc), metric2str(valid_loss, valid_acc), metric2str(test__loss, test__acc)) str2 = '{:14s} train : [{:}], valid : [{:}], test : [{:}]'.format(dataset, metric2str(train_info['loss'], train_info['accuracy']), metric2str(valid_info['loss'], valid_info['accuracy']), metric2str(test__info['loss'], test__info['accuracy']))
strings += [str1, str2] strings += [str1, str2]
if show: print('\n'.join(strings)) if show: print('\n'.join(strings))
return strings return strings
@ -34,19 +38,21 @@ class NASBench102API(object):
def __init__(self, file_path_or_dict, verbose=True): def __init__(self, file_path_or_dict, verbose=True):
if isinstance(file_path_or_dict, str): if isinstance(file_path_or_dict, str):
if verbose: print('try to create NAS-Bench-102 api from {:}'.format(file_path_or_dict)) if verbose: print('try to create the NAS-Bench-102 api from {:}'.format(file_path_or_dict))
assert os.path.isfile(file_path_or_dict), 'invalid path : {:}'.format(file_path_or_dict) assert os.path.isfile(file_path_or_dict), 'invalid path : {:}'.format(file_path_or_dict)
file_path_or_dict = torch.load(file_path_or_dict) file_path_or_dict = torch.load(file_path_or_dict)
else: else:
file_path_or_dict = copy.deepcopy( file_path_or_dict ) file_path_or_dict = copy.deepcopy( file_path_or_dict )
assert isinstance(file_path_or_dict, dict), 'It should be a dict instead of {:}'.format(type(file_path_or_dict)) assert isinstance(file_path_or_dict, dict), 'It should be a dict instead of {:}'.format(type(file_path_or_dict))
import pdb; pdb.set_trace() # we will update this api soon
keys = ('meta_archs', 'arch2infos', 'evaluated_indexes') keys = ('meta_archs', 'arch2infos', 'evaluated_indexes')
for key in keys: assert key in file_path_or_dict, 'Can not find key[{:}] in the dict'.format(key) for key in keys: assert key in file_path_or_dict, 'Can not find key[{:}] in the dict'.format(key)
self.meta_archs = copy.deepcopy( file_path_or_dict['meta_archs'] ) self.meta_archs = copy.deepcopy( file_path_or_dict['meta_archs'] )
self.arch2infos = OrderedDict() self.arch2infos_less = OrderedDict()
self.arch2infos_full = OrderedDict()
for xkey in sorted(list(file_path_or_dict['arch2infos'].keys())): for xkey in sorted(list(file_path_or_dict['arch2infos'].keys())):
self.arch2infos[xkey] = ArchResults.create_from_state_dict( file_path_or_dict['arch2infos'][xkey] ) all_info = file_path_or_dict['arch2infos'][xkey]
self.arch2infos_less[xkey] = ArchResults.create_from_state_dict( all_info['less'] )
self.arch2infos_full[xkey] = ArchResults.create_from_state_dict( all_info['full'] )
self.evaluated_indexes = sorted(list(file_path_or_dict['evaluated_indexes'])) self.evaluated_indexes = sorted(list(file_path_or_dict['evaluated_indexes']))
self.archstr2index = {} self.archstr2index = {}
for idx, arch in enumerate(self.meta_archs): for idx, arch in enumerate(self.meta_archs):
@ -73,35 +79,46 @@ class NASBench102API(object):
else: arch_index = -1 else: arch_index = -1
return arch_index return arch_index
def query_by_arch(self, arch): def query_by_arch(self, arch, use_12epochs_result=False):
if isinstance(arch, int):
arch_index = arch
else:
arch_index = self.query_index_by_arch(arch) arch_index = self.query_index_by_arch(arch)
if arch_index == -1: return None if arch_index == -1: return None # the following two lines are used to support few training epochs
if arch_index in self.arch2infos: if use_12epochs_result: arch2infos = self.arch2infos_less
strings = print_information(self.arch2infos[ arch_index ], 'arch-index={:}'.format(arch_index)) else : arch2infos = self.arch2infos_full
if arch_index in arch2infos:
strings = print_information(arch2infos[ arch_index ], 'arch-index={:}'.format(arch_index))
return '\n'.join(strings) return '\n'.join(strings)
else: else:
print ('Find this arch-index : {:}, but this arch is not evaluated.'.format(arch_index)) print ('Find this arch-index : {:}, but this arch is not evaluated.'.format(arch_index))
return None return None
def query_by_index(self, arch_index, dataname): def query_by_index(self, arch_index, dataname, use_12epochs_result=False):
assert arch_index in self.arch2infos, 'arch_index [{:}] does not in arch2info'.format(arch_index) if use_12epochs_result: basestr, arch2infos = '12epochs' , self.arch2infos_less
archInfo = copy.deepcopy( self.arch2infos[ arch_index ] ) else : basestr, arch2infos = '200epochs', self.arch2infos_full
assert arch_index in arch2infos, 'arch_index [{:}] does not in arch2info with {:}'.format(arch_index, basestr)
archInfo = copy.deepcopy( arch2infos[ arch_index ] )
assert dataname in archInfo.get_dataset_names(), 'invalid dataset-name : {:}'.format(dataname) assert dataname in archInfo.get_dataset_names(), 'invalid dataset-name : {:}'.format(dataname)
info = archInfo.query(dataname) info = archInfo.query(dataname)
return info return info
def query_meta_info_by_index(self, arch_index): def query_meta_info_by_index(self, arch_index, use_12epochs_result=False):
assert arch_index in self.arch2infos, 'arch_index [{:}] does not in arch2info'.format(arch_index) if use_12epochs_result: basestr, arch2infos = '12epochs' , self.arch2infos_less
archInfo = copy.deepcopy( self.arch2infos[ arch_index ] ) else : basestr, arch2infos = '200epochs', self.arch2infos_full
assert arch_index in arch2infos, 'arch_index [{:}] does not in arch2info with {:}'.format(arch_index, basestr)
archInfo = copy.deepcopy( arch2infos[ arch_index ] )
return archInfo return archInfo
def find_best(self, dataset, metric_on_set, FLOP_max=None, Param_max=None): def find_best(self, dataset, metric_on_set, FLOP_max=None, Param_max=None, use_12epochs_result=False):
if use_12epochs_result: basestr, arch2infos = '12epochs' , self.arch2infos_less
else : basestr, arch2infos = '200epochs', self.arch2infos_full
best_index, highest_accuracy = -1, None best_index, highest_accuracy = -1, None
for i, idx in enumerate(self.evaluated_indexes): for i, idx in enumerate(self.evaluated_indexes):
flop, param, latency = self.arch2infos[idx].get_comput_costs(dataset) flop, param, latency = arch2infos[idx].get_comput_costs(dataset)
if FLOP_max is not None and flop > FLOP_max : continue if FLOP_max is not None and flop > FLOP_max : continue
if Param_max is not None and param > Param_max: continue if Param_max is not None and param > Param_max: continue
loss, accuracy = self.arch2infos[idx].get_metrics(dataset, metric_on_set) loss, accuracy = arch2infos[idx].get_metrics(dataset, metric_on_set)
if best_index == -1: if best_index == -1:
best_index, highest_accuracy = idx, accuracy best_index, highest_accuracy = idx, accuracy
elif highest_accuracy < accuracy: elif highest_accuracy < accuracy:
@ -113,21 +130,29 @@ class NASBench102API(object):
return copy.deepcopy(self.meta_archs[index]) return copy.deepcopy(self.meta_archs[index])
def show(self, index=-1): def show(self, index=-1):
if index == -1: # show all architectures if index < 0: # show all architectures
print(self) print(self)
for i, idx in enumerate(self.evaluated_indexes): for i, idx in enumerate(self.evaluated_indexes):
print('\n' + '-' * 10 + ' The ({:5d}/{:5d}) {:06d}-th architecture! '.format(i, len(self.evaluated_indexes), idx) + '-'*10) print('\n' + '-' * 10 + ' The ({:5d}/{:5d}) {:06d}-th architecture! '.format(i, len(self.evaluated_indexes), idx) + '-'*10)
print('arch : {:}'.format(self.meta_archs[idx])) print('arch : {:}'.format(self.meta_archs[idx]))
strings = print_information(self.arch2infos[idx]) strings = print_information(self.arch2infos_full[idx])
print('>' * 20) print('>' * 40 + ' 200 epochs ' + '>' * 40)
print('\n'.join(strings)) print('\n'.join(strings))
print('<' * 20) strings = print_information(self.arch2infos_less[idx])
print('>' * 40 + ' 12 epochs ' + '>' * 40)
print('\n'.join(strings))
print('<' * 40 + '------------' + '<' * 40)
else: else:
if 0 <= index < len(self.meta_archs): if 0 <= index < len(self.meta_archs):
if index not in self.evaluated_indexes: print('The {:}-th architecture has not been evaluated or not saved.'.format(index)) if index not in self.evaluated_indexes: print('The {:}-th architecture has not been evaluated or not saved.'.format(index))
else: else:
strings = print_information(self.arch2infos[index]) strings = print_information(self.arch2infos_full[index])
print('>' * 40 + ' 200 epochs ' + '>' * 40)
print('\n'.join(strings)) print('\n'.join(strings))
strings = print_information(self.arch2infos_less[index])
print('>' * 40 + ' 12 epochs ' + '>' * 40)
print('\n'.join(strings))
print('<' * 40 + '------------' + '<' * 40)
else: else:
print('This index ({:}) is out of range (0~{:}).'.format(index, len(self.meta_archs))) print('This index ({:}) is out of range (0~{:}).'.format(index, len(self.meta_archs)))