def main(unused_argv):
set_random_seeds()
get_datapath() # The dataset path
get_steps() # setting steps according data_size
tf.logging.set_verbosity(tf.logging.INFO)
print('Now the mode of this mode is {} !'.format(FLAGS.mode))
# if log_dir is not exited, create it.
if not os.path.exists(FLAGS.log_dir): os.makedirs(FLAGS.log_dir)
if FLAGS.mode == 'decode':
FLAGS.branch_batch_size = FLAGS.beam_size # for beam search
FLAGS.TS_mode = False
hps = make_hps() # make a hps namedtuple
# Vocabulary
vocab = Vocab(hps.vocab_path, hps.vocab_size)
# Train or Inference
if hps.mode == 'train':
batcher = Batcher(hps.data_path, vocab, hps)
eval_hps = hps._replace(mode='eval')
eval_batcher = Batcher(hps.eval_data_path, vocab, eval_hps)
model = GSNModel(hps, vocab)
train(model, batcher, eval_batcher, vocab, hps)
elif hps.mode == 'decode':
decode_mdl_hps = hps._replace(max_dec_steps=1)
batcher = Batcher(hps.test_data_path, vocab, decode_mdl_hps) # for test
model = GSNModel(decode_mdl_hps, vocab)
decoder = BeamSearchDecoder(model, batcher, vocab)
decoder._decode()
def evaluation(request, *args, **kwargs):
if not request.session.get('is_login', False) or request.session['user_type'] != 'accounts':
return redirect("/index")
error_msg, page, page_list, projects_dict, project_id = initialize(args, request, True)
if not project_id:
return redirect('/account/summary')
eva_info = utils.get_eva_info(project_id)
steps = utils.get_steps(project_id)
step_path = utils.steps_path(utils.get_eva_info(project_id, top=True))
if request.method == 'GET':
return render(request, 'user_evaluation.html',
{"projects_dict": projects_dict, "project_id":project_id,
"eva_info": eva_info,
"steps": steps, "step_path": step_path,
"page_list":page_list, "page":page, "error_msg":error_msg})
if request.method == 'POST':
for num in [1,2,3,4,5]:
if request.FILES.get("file"+str(num)):
with open("database/evaluation.pk", 'rb') as f:
eva = pickle.load(f)
eva[project_id][num-1] = 1
with open("database/evaluation.pk", 'wb') as f:
f.write(pickle.dumps(eva))
utils.save_file(project_id, request, "file"+str(num), "evaluation%d"%num)
return redirect("/account/evaluation&id=%d&page=%d"%(project_id, page))
error_msg = '未选择文件'
return render(request, 'user_evaluation.html',
{"projects_dict": projects_dict, "project_id": project_id, "eva_info": eva_info,
'page_list': page_list, 'page': page, 'error_msg': error_msg})
def stop(request, *args, **kwargs):
if not request.session.get('is_login', False) or request.session['user_type'] != 'accounts':
return redirect("/index")
error_msg = ''
project_id = utils.get_id(args)
page = utils.get_page(args)
if request.method == "GET":
for id, item in utils.get_projects(request.session['account']):
if id == project_id:
project_info = item
return render(request, "user_projectstop.html", {"project_info":project_info, "project_id":project_id, "page":page , "error_msg": error_msg})
if request.method == "POST":
data = {
"name": request.POST.get("name"),
"leader": request.POST.get("leader"),
"teammates": request.POST.get("teammates"),
"start": request.POST.get("start"),
"cost": request.POST.get("cost"),
}
if int(data['cost']) < 0:
data['cost'] = '0'
if not error_msg:
error_msg = "项目申停成功"
_projects_dict = utils.get_projects(request.session['account'])
page_list = utils.get_pages(page, _projects_dict)
projects_dict, page, project_id = utils.adjust_info(_projects_dict, page, project_id)
steps = utils.get_steps(project_id)
step_path = utils.steps_path(utils.get_eva_info(project_id, top=True))
return render(request, "user_projectprocess.html",
{"projects_dict": projects_dict, "project_id": project_id,
"steps": steps, "step_path": step_path, "page_list": page_list, "page": page,
'error_msg': error_msg})
def process(request, *args, **kwargs):
if not request.session.get('is_login', False) or request.session['user_type'] != 'accounts':
return redirect("/index")
error_msg, page, page_list, projects_dict, project_id = initialize(args, request, True)
if not project_id:
return redirect('/account/summary')
steps = utils.get_steps(project_id)
step_path = utils.steps_path(utils.get_eva_info(project_id, top=True))
if request.method == 'GET':
return render(request, 'user_projectprocess.html',{"projects_dict": projects_dict, "project_id":project_id,
"steps":steps, "step_path":step_path, "page_list":page_list, "page":page, 'error_msg':error_msg})
if request.method == "POST":
data = {
"mission" : request.POST.get("mission"),
"flow": request.POST.get("flow"),
"start" : request.POST.get("start"),
"process" : request.POST.get("process"),
"teammates": request.POST.get("teammates").strip().split(',') if request.POST.get("teammates") else None,
"end" : request.POST.get("end"),
}
data, error_msg = utils.check_error(data, pro_name=False, budget=False)
if not data.get("mission", None):
error_msg = '请输入正确的任务名称'
elif data.get("flow", None) not in ['1','2','3','4','5']:
error_msg = '请输入正确的任务所属流程'
# utils.add_steps(project_id, data)
# return redirect("/account/process&id=%d&page=%d"%(project_id, page))
if not error_msg:
utils.add_steps(project_id, data)
return redirect("/account/process&id=%d&page=%d" % (project_id, page))
else:
return render(request, 'user_projectprocess.html',
{"projects_dict": projects_dict, "project_id": project_id,
"steps": steps, "step_path": step_path, "page_list": page_list, "page": page,
'error_msg': error_msg})
def main(queue, name):
import keras
import utils
import model as MyModels
assert name + '_model' in MyModels.__dict__.keys()
config = utils.MyConfig(type=name, train_epochs=1000, train_batch_size=16)
model = MyModels.__dict__[name + '_model'](input_shape=(256, 256) +
(config.input_channels, ))
try:
model.load_weights(config.model_path, by_name=True)
pass
except Exception as inst:
print inst
# exit(-2)
model.summary()
callback_list = [
keras.callbacks.ModelCheckpoint(config.model_path,
monitor='loss2acc',
save_best_only=True,
mode='max',
save_weights_only=False),
keras.callbacks.TensorBoard(log_dir='tf_tmp/')
]
my_metric = lambda x, y: MyModels.loss2acc(x, y, True)
my_metric.__name__ = 'loss2acc'
model.compile(optimizer=keras.optimizers.adam(lr=1e-3),
loss=['mse'],
metrics=[my_metric])
dbg = False
model.fit_generator(
utils.gen_from_dir(config, mode=True),
steps_per_epoch=1 if dbg else utils.get_steps(config, train=True),
epochs=2 if dbg else config.train_epochs,
callbacks=callback_list,
validation_steps=utils.get_steps(config, train=False),
validation_data=utils.gen_from_dir(config, mode=False))
# model.save(config.model_path)
queue.put({'model_path': config.model_path})
def process(request, *args, **kwargs):
if not request.session.get(
'is_login', False) or request.session['user_type'] != 'admin':
return redirect("/index")
error_msg, page, page_list, projects_dict, project_id = initialize(
args, request)
steps = utils.get_steps(project_id)
step_list = []
for obj in steps:
step_list.append(obj[1]['flow'])
step_path = utils.steps_path(utils.get_eva_info(project_id, top=True))
if request.method == 'GET':
return render(
request, 'admin_projectprocess.html', {
"projects_dict": projects_dict,
"project_id": project_id,
"steps": steps,
"step_path": step_path,
"page_list": page_list,
"page": page,
'error_msg': error_msg
})
if request.method == "POST":
data = {
"mission": request.POST.get("mission"),
"flow": request.POST.get("flow"),
"leader": request.POST.get("leader"),
"start": request.POST.get("start"),
"process": request.POST.get("process"),
"teammates": request.POST.get("teammates"),
"end": request.POST.get("end"),
}
if not data.get("mission", None):
error_msg = '请输入正确的任务名称'
elif data.get("flow", None) not in ['1', '2', '3', '4', '5']:
error_msg = '请输入正确的任务所属流程'
# utils.add_steps(project_id, data)
# return redirect("/account/process&id=%d&page=%d"%(project_id, page))
if not error_msg:
utils.add_steps(project_id, data)
return redirect("/admin/process&id=%d&page=%d" %
(project_id, page))
else:
return render(
request, 'admin_projcetprocess.html', {
"projects_dict": projects_dict,
"project_id": project_id,
"steps": steps,
"step_path": step_path,
"page_list": page_list,
"page": page,
'error_msg': error_msg
})
def finance(request, *args, **kwargs):
if not request.session.get('is_login', False) or request.session['user_type'] != 'accounts':
return redirect("/index")
error_msg, page, page_list, projects_dict, project_id = initialize(args, request, True)
if not project_id:
return redirect('/account/summary')
steps = utils.get_steps(project_id)
step_path = utils.steps_path(utils.get_eva_info(project_id, top=True))
if request.method == 'GET':
with open("database/finance.pk", 'rb') as f:
finance = pickle.load(f)
records = enumerate(finance.get(project_id, []))
return render(request, 'user_finance.html', {"records":records,"projects_dict": projects_dict, "project_id":project_id,
"steps": steps, "step_path": step_path,"page_list":page_list, "page":page, "error_msg":error_msg})
def train_albert(train_df, val_df, fold_i, augment=False):
max_l = Config.Albert.max_len
_train_generator = AlbertDataGenerator(train_df, augment=augment)
train_dataset = tf.data.Dataset.from_generator(_train_generator.generate,
output_types=({
'ids':
tf.int32,
'att':
tf.int32,
'tti':
tf.int32
}, {
'sts':
tf.int32,
'ets':
tf.int32
}))
train_dataset = train_dataset.padded_batch(Config.Train.batch_size,
padded_shapes=({
'ids': [None],
'att': [None],
'tti': [None]
}, {
'sts': [None],
'ets': [None]
}))
train_dataset = train_dataset.repeat().prefetch(
tf.data.experimental.AUTOTUNE)
_val_generator = AlbertDataGenerator(val_df, augment=False)
val_dataset = tf.data.Dataset.from_generator(_val_generator.generate,
output_types=({
'ids': tf.int32,
'att': tf.int32,
'tti': tf.int32
}, {
'sts': tf.int32,
'ets': tf.int32
}))
val_dataset = val_dataset.padded_batch(Config.Train.batch_size,
padded_shapes=({
'ids': [None],
'att': [None],
'tti': [None]
}, {
'sts': [None],
'ets': [None]
}))
val_dataset = val_dataset.repeat().prefetch(tf.data.experimental.AUTOTUNE)
model = get_albert()
if fold_i == 0:
model.summary()
model_name = f'weights_v{Config.version}_f{fold_i + 1}.h5'
train_spe = get_steps(train_df)
val_spe = get_steps(val_df)
cbs = [
WarmUpCosineDecayScheduler(6e-5,
1200,
warmup_steps=300,
hold_base_rate_steps=200,
verbose=0),
keras.callbacks.ModelCheckpoint(str(Config.Train.checkpoint_dir /
Config.model_type / model_name),
verbose=1,
save_best_only=True,
save_weights_only=True)
]
model.fit(train_dataset,
epochs=2,
verbose=1,
validation_data=val_dataset,
callbacks=cbs,
steps_per_epoch=train_spe,
validation_steps=val_spe)
print(f'Loading checkpoint {model_name}...')
model.load_weights(
str(Config.Train.checkpoint_dir / Config.model_type / model_name))
_val_generator = AlbertDataGenerator(val_df, augment=False)
val_dataset = tf.data.Dataset.from_generator(_val_generator.generate,
output_types=({
'ids': tf.int32,
'att': tf.int32,
'tti': tf.int32
}, {
'sts': tf.int32,
'ets': tf.int32
}))
val_dataset = val_dataset.padded_batch(Config.Train.batch_size,
padded_shapes=({
'ids': [max_l],
'att': [max_l],
'tti': [max_l]
}, {
'sts': [max_l],
'ets': [max_l]
}))
val_dataset = val_dataset.prefetch(tf.data.experimental.AUTOTUNE)
s_idx, e_idx = model.predict(val_dataset, verbose=1)
s_idx = np.argmax(s_idx, axis=-1)
e_idx = np.argmax(e_idx, axis=-1)
jaccard_score = get_jaccard_from_df(val_df, s_idx, e_idx, 'albert',
'albert.csv')
print(
f'\n>>> Fold {fold_i + 1}: jaccard_score for albert: {jaccard_score}\n'
)
return jaccard_score