fix train module

deepsignal_plant/call_mods_freq.py

@@ -20,6 +20,8 @@
 from .utils.process_utils import is_file_empty
 import uuid
 
+os.environ['MKL_THREADING_LAYER'] = 'GNU'
+
 time_wait = 3
 
 

deepsignal_plant/dataloader.py

@@ -33,9 +33,9 @@ def parse_a_line2(line):
 
 class SignalFeaData2(Dataset):
  def __init__(self, filename, transform=None):
- print(">>>using linecache to access '{}'<<<\n"
- ">>>after done using the file, "
- "remember to use linecache.clearcache() to clear cache for safety<<<".format(filename))
+ # print(">>>using linecache to access '{}'<<<\n"
+ #  ">>>after done using the file, "
+ #  "remember to use linecache.clearcache() to clear cache for safety<<<".format(filename))
  self._filename = os.path.abspath(filename)
  self._total_data = 0
  self._transform = transform

deepsignal_plant/deepsignal_plant.py

@@ -55,9 +55,8 @@ def main_call_mods(args):
 
 
 def main_call_freq(args):
- import os
- os.environ['MKL_THREADING_LAYER'] = 'GNU'
  from .call_mods_freq import call_mods_frequency_to_file
+
  display_args(args)
  call_mods_frequency_to_file(args)
 

deepsignal_plant/train.py

@@ -103,6 +103,7 @@ def train(args):
  # train at most max_epoch_num epochs
  for epoch in range(args.max_epoch_num):
  curr_best_accuracy_epoch = 0
+ no_best_model = True
  tlosses = []
  start = time.time()
  for i, sfeatures in enumerate(train_loader):
@@ -147,13 +148,6 @@ def train(args):
  if use_cuda:
  vlabels = vlabels.cpu()
  vpredicted = vpredicted.cpu()
- # i_accuracy = metrics.accuracy_score(vlabels.numpy(), vpredicted)
- # i_precision = metrics.precision_score(vlabels.numpy(), vpredicted)
- # i_recall = metrics.recall_score(vlabels.numpy(), vpredicted)
-
- # vaccus.append(i_accuracy)
- # vprecs.append(i_precision)
- # vrecas.append(i_recall)
  vlosses.append(vloss.item())
  vlabels_total += vlabels.tolist()
  vpredicted_total += vpredicted.tolist()
@@ -163,11 +157,14 @@ def train(args):
  v_recall = metrics.recall_score(vlabels_total, vpredicted_total)
  if v_accuracy > curr_best_accuracy_epoch:
  curr_best_accuracy_epoch = v_accuracy
- if curr_best_accuracy_epoch > curr_best_accuracy - 0.0005:
+ if curr_best_accuracy_epoch > curr_best_accuracy - 0.0002:
  torch.save(model.state_dict(),
  model_dir + args.model_type + '.b{}_s{}_epoch{}.ckpt'.format(args.seq_len,
  args.signal_len,
  epoch + 1))
+ if curr_best_accuracy_epoch > curr_best_accuracy:
+ curr_best_accuracy = curr_best_accuracy_epoch
+ no_best_model = False
 
  time_cost = time.time() - start
  print('Epoch [{}/{}], Step [{}/{}], TrainLoss: {:.4f}; '
@@ -182,16 +179,16 @@ def train(args):
  sys.stdout.flush()
  model.train()
  scheduler.step()
- if curr_best_accuracy_epoch > curr_best_accuracy:
- curr_best_accuracy = curr_best_accuracy_epoch
- else:
- if epoch >= args.min_epoch_num - 1:
- print("best accuracy: {}, early stop!".format(curr_best_accuracy))
- break
+
+ if no_best_model and epoch >= args.min_epoch_num - 1:
+ print("early stop!")
+ break
 
  endtime = time.time()
  clear_linecache()
- print("[train] training cost {} seconds".format(endtime - total_start))
+ print("[train]training cost {} seconds, "
+ "best accuracy: {}".format(endtime - total_start, 
+ curr_best_accuracy))
 
 
 def main():