def args_to_dicts(args):
global parse_args
print(args)
algorithm = args['algorithm_type']
algorithm_type = algorithm['type']
#print(algorithm_type)
algorithms = {}
_LOG_REG = False
_XGB = False
_SVM = False
if algorithm_type == 'logreg':
_LOG_REG = True
algorithms['log_reg'] = algorithm
elif algorithm_type == 'svm':
_SVM = True
algorithms['svm'] = algorithm
else:
_XGB = True
algorithms['xgb'] = algorithm
passed_args = copy.deepcopy(args)
passed_args.pop('algorithm_type')
passed_args['RESAMPLING'] = True
passed_args['USE_CACHE'] = True
passed_args['W2V_PRETRAINED'] = True
#print(_LOG_REG, _SVM, _XGB)
# Because the algoritm is above we don't actually need it to be set here
directory, features, _, parameters = mp.get_args(
directory=parse_args.directory_base,
LOG_REG=_LOG_REG,
SVM=_SVM,
XGB=_XGB,
**passed_args)
return directory, features, algorithms, parameters
def run_experiment(
directory, BOW_APPEND, BOW_DIFFERENCE, BOW_PRODUCT, BOW_COS, BOW_TFIDF,
ST_APPEND, ST_DIFFERENCE, ST_PRODUCT, ST_COS, LDA_APPEND,
LDA_DIFFERENCE, LDA_PRODUCT, LDA_COS, LDA_TOPICS, W2V_AVG, W2V_MAX,
W2V_MIN, W2V_ABS, W2V_APPEND, W2V_DIFFERENCE, W2V_PRODUCT, W2V_COS,
W2V_PRETRAINED, W2V_MIN_COUNT, W2V_WINDOW, W2V_SIZE, W2V_WORKERS,
CNN_APPEND, CNN_DIFFERENCE, CNN_PRODUCT, CNN_COS, WORDONEHOT,
WORDONEHOT_VOCAB, LOG_REG, LOG_PENALTY, LOG_TOL, LOG_C, SVM, SVM_C,
SVM_KERNEL, SVM_GAMMA, XGB, XGB_LEARNRATE, XGB_MAXDEPTH,
XGB_MINCHILDWEIGHT, XGB_COLSAMPLEBYTREE, MEM_NET, MEM_VOCAB, MEM_TYPE,
MEM_BATCH, MEM_EPOCHS, MEM_MASK_MODE, MEM_EMBED_MODE,
MEM_ONEHOT_MIN_LEN, MEM_ONEHOT_MAX_LEN, RESAMPLING, NOVEL_RATIO,
OVERSAMPLING, REPLACEMENT, SAVE_RESULTS, SAVEEXPERIMENTDATA,
EXPERIMENTDATAFILE, VOCAB_SIZE, STEM, FULL_VOCAB_SIZE, FULL_VOCAB_TYPE,
FULL_CHAR_VOCAB, SEED, USE_CACHE):
return pythia_main(
get_args(directory, BOW_APPEND, BOW_DIFFERENCE, BOW_PRODUCT, BOW_COS,
BOW_TFIDF, ST_APPEND, ST_DIFFERENCE, ST_PRODUCT, ST_COS,
LDA_APPEND, LDA_DIFFERENCE, LDA_PRODUCT, LDA_COS, LDA_TOPICS,
W2V_AVG, W2V_MAX, W2V_MIN, W2V_ABS, W2V_APPEND,
W2V_DIFFERENCE, W2V_PRODUCT, W2V_COS, W2V_PRETRAINED,
W2V_MIN_COUNT, W2V_WINDOW, W2V_SIZE, W2V_WORKERS, CNN_APPEND,
CNN_DIFFERENCE, CNN_PRODUCT, CNN_COS, WORDONEHOT,
WORDONEHOT_VOCAB, LOG_REG, LOG_PENALTY, LOG_TOL, LOG_C, SVM,
SVM_C, SVM_KERNEL, SVM_GAMMA, XGB, XGB_LEARNRATE,
XGB_MAXDEPTH, XGB_MINCHILDWEIGHT, XGB_COLSAMPLEBYTREE,
MEM_NET, MEM_VOCAB, MEM_TYPE, MEM_BATCH, MEM_EPOCHS,
MEM_MASK_MODE, MEM_EMBED_MODE, MEM_ONEHOT_MIN_LEN,
MEM_ONEHOT_MAX_LEN, RESAMPLING, NOVEL_RATIO, OVERSAMPLING,
REPLACEMENT, SAVE_RESULTS, SAVEEXPERIMENTDATA,
EXPERIMENTDATAFILE, VOCAB_SIZE, STEM, FULL_VOCAB_SIZE,
FULL_VOCAB_TYPE, FULL_CHAR_VOCAB, SEED, USE_CACHE))
def args_to_dicts(args):
global parse_args
print(args)
algorithm= args['algorithm_type']
algorithm_type = algorithm['type']
#print(algorithm_type)
algorithms = {}
_LOG_REG = False
_XGB = False
_SVM = False
if algorithm_type=='logreg':
_LOG_REG = True
algorithms['log_reg'] = algorithm
elif algorithm_type== 'svm':
_SVM = True
algorithms['svm'] = algorithm
else:
_XGB = True
algorithms['xgb'] = algorithm
passed_args = copy.deepcopy(args)
passed_args.pop('algorithm_type')
passed_args['RESAMPLING']=True
passed_args['USE_CACHE']=True
passed_args['W2V_PRETRAINED']=True
#print(_LOG_REG, _SVM, _XGB)
# Because the algoritm is above we don't actually need it to be set here
directory, features, _, parameters = mp.get_args(
directory=parse_args.directory_base, LOG_REG=_LOG_REG, SVM=_SVM, XGB=_XGB, **passed_args)
return directory, features, algorithms, parameters
def objective(args_):
# arguments to pass as config_updates dict
global args
# result to pass to hyperopt
global result
# command-line arguments
global parse_args
try:
ex = Experiment('Hyperopt')
logger.debug("Adding observer for {}, DB {}".format(parse_args.mongo_db_address,parse_args.mongo_db_name))
ex.observers.append(MongoObserver.create(url=parse_args.mongo_db_address, db_name=parse_args.mongo_db_name))
pythia_args = make_args_for_pythia(args_)
args = mp.get_args(**pythia_args)
ex.main(run_with_global_args)
r = ex.run(config_updates=pythia_args)
logger.debug("Experiment result: {}\n"
"Report to hyperopt: {}".format(r.result, result))
return result
except:
raise
#If we somehow cannot get to the MongoDB server, then continue with the experiment
logger.warning("Running without Sacred")
run_with_global_args()
def objective(args_):
# arguments to pass as config_updates dict
global args
# result to pass to hyperopt
global result
# command-line arguments
global parse_args
try:
ex = Experiment('Hyperopt')
logger.debug("Adding observer for {}, DB {}".format(
parse_args.mongo_db_address, parse_args.mongo_db_name))
ex.observers.append(
MongoObserver.create(url=parse_args.mongo_db_address,
db_name=parse_args.mongo_db_name))
pythia_args = make_args_for_pythia(args_)
args = mp.get_args(**pythia_args)
ex.main(run_with_global_args)
r = ex.run(config_updates=pythia_args)
logger.debug("Experiment result: {}\n"
"Report to hyperopt: {}".format(r.result, result))
return result
except:
raise
#If we somehow cannot get to the MongoDB server, then continue with the experiment
logger.warning("Running without Sacred")
run_with_global_args()
def test_random():
args = get_args(SEED=41,
directory='data/stackexchange/anime',
XGB=True,
LDA_APPEND=True,
BOW_APPEND=True,
RESAMPLING=False)
run1 = pipeline_main(args)
run2 = pipeline_main(args)
assert run1==run2
def run_experiment(
directory, BOW_APPEND, BOW_DIFFERENCE, BOW_PRODUCT, BOW_COS, BOW_TFIDF,
BOW_BINARY, ST_APPEND, ST_DIFFERENCE, ST_PRODUCT, ST_COS, LDA_APPEND,
LDA_DIFFERENCE, LDA_PRODUCT, LDA_COS, LDA_TOPICS, W2V_AVG, W2V_MAX,
W2V_MIN, W2V_ABS, W2V_APPEND, W2V_DIFFERENCE, W2V_PRODUCT, W2V_COS,
W2V_PRETRAINED, W2V_MIN_COUNT, W2V_WINDOW, W2V_SIZE, W2V_WORKERS,
CNN_APPEND, CNN_DIFFERENCE, CNN_PRODUCT, CNN_COS, WORDONEHOT,
WORDONEHOT_VOCAB, LOG_REG, LOG_PENALTY, LOG_TOL, LOG_C, SVM, SVM_C,
SVM_KERNEL, SVM_GAMMA, XGB, XGB_LEARNRATE, XGB_MAXDEPTH,
XGB_MINCHILDWEIGHT, XGB_COLSAMPLEBYTREE, SGD, SGD_LOSS, SGD_ALPHA,
SGD_PENALTY, SGD_EPOCHS, SGD_BATCH_SIZE, MEM_NET, MEM_VOCAB, MEM_TYPE,
MEM_BATCH, MEM_EPOCHS, MEM_MASK_MODE, MEM_EMBED_MODE,
MEM_ONEHOT_MIN_LEN, MEM_ONEHOT_MAX_LEN, RESAMPLING, NOVEL_RATIO,
OVERSAMPLING, REPLACEMENT, SAVE_RESULTS, SAVEEXPERIMENTDATA,
EXPERIMENTDATAFILE, VOCAB_SIZE, STEM, FULL_VOCAB_SIZE, FULL_VOCAB_TYPE,
FULL_CHAR_VOCAB, FULL_VOCAB_STEM, SEED, HDF5_PATH_TRAIN,
HDF5_PATH_TEST, HDF5_SAVE_FREQUENCY, HDF5_USE_EXISTING, USE_CACHE,
_run):
# store default metadata
USER = os.environ.get('USER', 'unknown user')
_run.info = {'user': USER}
return pythia_main(
get_args(directory, BOW_APPEND, BOW_DIFFERENCE, BOW_PRODUCT, BOW_COS,
BOW_TFIDF, BOW_BINARY, ST_APPEND, ST_DIFFERENCE, ST_PRODUCT,
ST_COS, LDA_APPEND, LDA_DIFFERENCE, LDA_PRODUCT, LDA_COS,
LDA_TOPICS, W2V_AVG, W2V_MAX, W2V_MIN, W2V_ABS, W2V_APPEND,
W2V_DIFFERENCE, W2V_PRODUCT, W2V_COS, W2V_PRETRAINED,
W2V_MIN_COUNT, W2V_WINDOW, W2V_SIZE, W2V_WORKERS, CNN_APPEND,
CNN_DIFFERENCE, CNN_PRODUCT, CNN_COS, WORDONEHOT,
WORDONEHOT_VOCAB, LOG_REG, LOG_PENALTY, LOG_TOL, LOG_C, SVM,
SVM_C, SVM_KERNEL, SVM_GAMMA, XGB, XGB_LEARNRATE,
XGB_MAXDEPTH, XGB_MINCHILDWEIGHT, XGB_COLSAMPLEBYTREE, SGD,
SGD_LOSS, SGD_ALPHA, SGD_PENALTY, SGD_EPOCHS, SGD_BATCH_SIZE,
MEM_NET, MEM_VOCAB, MEM_TYPE, MEM_BATCH, MEM_EPOCHS,
MEM_MASK_MODE, MEM_EMBED_MODE, MEM_ONEHOT_MIN_LEN,
MEM_ONEHOT_MAX_LEN, RESAMPLING, NOVEL_RATIO, OVERSAMPLING,
REPLACEMENT, SAVE_RESULTS, SAVEEXPERIMENTDATA,
EXPERIMENTDATAFILE, VOCAB_SIZE, STEM, FULL_VOCAB_SIZE,
FULL_VOCAB_TYPE, FULL_CHAR_VOCAB, FULL_VOCAB_STEM, SEED,
HDF5_PATH_TRAIN, HDF5_PATH_TEST, HDF5_SAVE_FREQUENCY,
HDF5_USE_EXISTING, USE_CACHE))
def run_experiment(
directory,
BOW_APPEND,
BOW_DIFFERENCE,
BOW_PRODUCT,
BOW_COS,
BOW_TFIDF,
BOW_VOCAB,
ST_APPEND,
ST_DIFFERENCE,
ST_PRODUCT,
ST_COS,
LDA_APPEND,
LDA_DIFFERENCE,
LDA_PRODUCT,
LDA_COS,
LDA_VOCAB,
LDA_TOPICS,
LOG_REG,
LOG_PENALTY,
LOG_TOL,
LOG_C,
SVM,
SVM_C,
SVM_KERNAL,
SVM_GAMMA,
XGB,
XGB_LEARNRATE,
XGB_MAXDEPTH,
XGB_MINCHILDWEIGHT,
XGB_COLSAMPLEBYTREE,
RESAMPLING,
NOVEL_RATIO,
OVERSAMPLING,
REPLACEMENT,
SEED):
return pythia_main(
get_args(
directory,
BOW_APPEND,
BOW_DIFFERENCE,
BOW_PRODUCT,
BOW_COS,
BOW_TFIDF,
BOW_VOCAB,
ST_APPEND,
ST_DIFFERENCE,
ST_PRODUCT,
ST_COS,
LDA_APPEND,
LDA_DIFFERENCE,
LDA_PRODUCT,
LDA_COS,
LDA_VOCAB,
LDA_TOPICS,
LOG_REG,
LOG_PENALTY,
LOG_TOL,
LOG_C,
SVM,
SVM_C,
SVM_KERNAL,
SVM_GAMMA,
XGB,
XGB_LEARNRATE,
XGB_MAXDEPTH,
XGB_MINCHILDWEIGHT,
XGB_COLSAMPLEBYTREE,
RESAMPLING,
NOVEL_RATIO,
OVERSAMPLING,
REPLACEMENT,
SEED)
)
def run_experiment(directory,
BOW_APPEND,
BOW_DIFFERENCE,
BOW_PRODUCT,
BOW_COS,
BOW_TFIDF,
ST_APPEND,
ST_DIFFERENCE,
ST_PRODUCT,
ST_COS,
LDA_APPEND,
LDA_DIFFERENCE,
LDA_PRODUCT,
LDA_COS,
LDA_TOPICS,
W2V_AVG,
W2V_MAX,
W2V_MIN,
W2V_ABS,
W2V_APPEND,
W2V_DIFFERENCE,
W2V_PRODUCT,
W2V_COS,
W2V_PRETRAINED,
W2V_MIN_COUNT,
W2V_WINDOW,
W2V_SIZE,
W2V_WORKERS,
CNN_APPEND,
CNN_DIFFERENCE,
CNN_PRODUCT,
CNN_COS,
WORDONEHOT,
WORDONEHOT_VOCAB,
LOG_REG,
LOG_PENALTY,
LOG_TOL,
LOG_C,
SVM,
SVM_C,
SVM_KERNEL,
SVM_GAMMA,
XGB,
XGB_LEARNRATE,
XGB_MAXDEPTH,
XGB_MINCHILDWEIGHT,
XGB_COLSAMPLEBYTREE,
MEM_NET,
MEM_VOCAB,
MEM_TYPE,
MEM_BATCH,
MEM_EPOCHS,
MEM_MASK_MODE,
MEM_EMBED_MODE,
MEM_ONEHOT_MIN_LEN,
MEM_ONEHOT_MAX_LEN,
RESAMPLING,
NOVEL_RATIO,
OVERSAMPLING,
REPLACEMENT,
SAVE_RESULTS,
SAVEEXPERIMENTDATA,
EXPERIMENTDATAFILE,
VOCAB_SIZE,
STEM,
FULL_VOCAB_SIZE,
FULL_VOCAB_TYPE,
FULL_CHAR_VOCAB,
SEED,
USE_CACHE):
return pythia_main(
get_args(
directory,
BOW_APPEND,
BOW_DIFFERENCE,
BOW_PRODUCT,
BOW_COS,
BOW_TFIDF,
ST_APPEND,
ST_DIFFERENCE,
ST_PRODUCT,
ST_COS,
LDA_APPEND,
LDA_DIFFERENCE,
LDA_PRODUCT,
LDA_COS,
LDA_TOPICS,
W2V_AVG,
W2V_MAX,
W2V_MIN,
W2V_ABS,
W2V_APPEND,
W2V_DIFFERENCE,
W2V_PRODUCT,
W2V_COS,
W2V_PRETRAINED,
W2V_MIN_COUNT,
W2V_WINDOW,
W2V_SIZE,
W2V_WORKERS,
CNN_APPEND,
CNN_DIFFERENCE,
CNN_PRODUCT,
CNN_COS,
WORDONEHOT,
WORDONEHOT_VOCAB,
LOG_REG,
LOG_PENALTY,
LOG_TOL,
LOG_C,
SVM,
SVM_C,
SVM_KERNEL,
SVM_GAMMA,
XGB,
XGB_LEARNRATE,
XGB_MAXDEPTH,
XGB_MINCHILDWEIGHT,
XGB_COLSAMPLEBYTREE,
MEM_NET,
MEM_VOCAB,
MEM_TYPE,
MEM_BATCH,
MEM_EPOCHS,
MEM_MASK_MODE,
MEM_EMBED_MODE,
MEM_ONEHOT_MIN_LEN,
MEM_ONEHOT_MAX_LEN,
RESAMPLING,
NOVEL_RATIO,
OVERSAMPLING,
REPLACEMENT,
SAVE_RESULTS,
SAVEEXPERIMENTDATA,
EXPERIMENTDATAFILE,
VOCAB_SIZE,
STEM,
FULL_VOCAB_SIZE,
FULL_VOCAB_TYPE,
FULL_CHAR_VOCAB,
SEED,
USE_CACHE)
)
