action='store',
dest='num_epochs',
type=int,
default=2)
parser.add_argument('--gpu', action="store_true", default=False)
results = parser.parse_args()
data_dir_in = results.data_dir
arch_in = results.arch
save_dir_in = results.save_dir
lr_in = results.lr
hidden_size = results.h_units
epochs = results.num_epochs
gpu_on = results.gpu
train_data, trainloader, validloader, testloader = load_data(data_dir_in)
model = getattr(models, arch_in)(pretrained=True)
for param in model.parameters():
param.requires_grad = False
build_model(model, hidden_size)
train_model(model, gpu_on, trainloader, validloader, lr_in, epochs)
test_model(model, gpu_on, testloader)
save_model(model, arch_in, train_data, save_dir_in)
def api_keys():
api_keys = load_data('api_keys.yaml')
return render_template('api_keys.html',api_keys=api_keys)
def api_docs():
api = load_data('api.yaml')
return render_template('api_docs.html',api=api)
from keras.models import Sequential
from keras.layers import Dense
import numpy as np
import funcs
from collections import Counter
if __name__ == '__main__':
print('Getting data')
# data_train, label_train = funcs.get_train_data()
data_train = funcs.load_data("INPUT_DATA.pkl")
label_train = funcs.load_data("INPUT_LABELS.pkl")
print("len of train " + str(len(data_train)))
print("len of lable " + str(len(label_train)))
print("===========================")
# # Provide Input Data
# # ========================================================
# input_data = []
# for data_item in data_train:
# mean = np.mean(data_item)
#
# dt = Counter(data_item)
# ent = np.zeros(129)
# ent[0] = mean
# for idx in range(1,128):
# ent[idx] = dt[idx]
# input_data.append(ent)
# input_data = np.asarray(input_data)
# # print(len(input_data))
# # print(input_data[0])
# # exit()
bin_width_ms = 1
bin_width_ms_session = 50
smooth_sd_ms = 100
fr_convert = 1000
trial_duration = 2000
bin_edges_trial = np.arange(0 ,trial_duration, bin_width_ms)
max_list = []
cluster_list = []
start_trial_list = []
end_trial_list = []
spikes_times_session_list =[]
for recording_to_extract,session in zip(recordings_to_extract_hp,sessions_hp):cluster_list
path_to_data = '/'.join([kilosort_folder, recording_to_extract])
os.chdir(path_to_data)
ephys_session = fu.load_data(recording_to_extract,kilosort_folder,'/',True )
beh_session = di.Session('/media/behrenslab/90c50efc-05cf-4045-95e4-9dabd129fb47/Ephys_Reversal_Learning/data/Reversal_learning Behaviour Data and Code/data_3_tasks_ephys/{}'.format(session))
forced_trials = beh_session.trial_data['forced_trial']
non_forced_array = np.where(forced_trials == 0)[0]
task = beh_session.trial_data['task']
task_non_forced = task[non_forced_array]
#Trial Initiation Timestamps
pyControl_choice = [event.time for event in beh_session.events if event.name in ['choice_state']]
pyControl_choice = np.array(pyControl_choice)
pyControl_end_trial = [event.time for event in beh_session.events if event.name in ['inter_trial_interval']][2:] #first two ITIs are free rewards
pyControl_end_trial = np.array(pyControl_end_trial)
task = beh_session.trial_data['task']
task_1_end_trial = np.where(task == 1)[0]
task_2_end_trial = np.where(task == 2)[0]
task_2_change = np.where(task ==2)[0]
table.write(4, 6, 'k_minus:' + str(k_alpha_minus))
table.write(5, 6, 'PINC:' + str(PINC))
# clear current error times
current_error_times = 0
# select data set
[file_name, sheet_name, col, row_start, seq_len,
interval] = dataset.select_dataset_windpower8_boundvmd(flag)
# preprocess data
print('> Loading data... ')
[X_train, Y_train, X_test, Y_test] = fc.load_data(file_name,
type='excel',
sheet_name=sheet_name,
pre_seq_len=timesteps,
row=row_start - 1,
col=col - 1,
seq_len=seq_len,
interval=interval)
# normalization
[x_train, x_maxmin] = fc.maponezero(X_train)
[y_train, y_maxmin] = fc.maponezero(Y_train)
x_test = fc.maponezero(X_test, "apply", x_maxmin)
y_test = fc.maponezero(Y_test, "apply", y_maxmin)
x_train = np.reshape(x_train, (x_train.shape[0], x_train.shape[1], 1))
x_test = np.reshape(x_test, (x_test.shape[0], x_test.shape[1], 1))
for times in range(0, exp_times):
while True:
width_upper_history = []
width_lower_history = []
alpha_history = []
