解决 -bash: ./experiment-rs.sh: /bin/bash^M: 解释器错误: 没有那个文件或目录

src/learn_framework.py

@@ -76,13 +76,13 @@ def run_train(self, train_data, dev_data):
 
         for epoch_id in range(self.start_epoch, self.num_epochs):
             print('Epoch {}'.format(epoch_id))
-            if self.rl_variation_tag.startswith('rs'):
-                # Reward shaping module sanity check:
-                #   Make sure the reward shaping module output value is in the correct range
-                train_scores = self.test_fn(train_data)
-                dev_scores = self.test_fn(dev_data)
-                print('Train set average fact score: {}'.format(float(train_scores.mean())))
-                print('Dev set average fact score: {}'.format(float(dev_scores.mean())))
+            # if self.rl_variation_tag.startswith('rs'):
+            #     # Reward shaping module sanity check:
+            #     #   Make sure the reward shaping module output value is in the correct range
+            #     train_scores = self.test_fn(train_data)
+            #     dev_scores = self.test_fn(dev_data)
+            #     print('Train set average fact score: {}'.format(float(train_scores.mean())))
+            #     print('Dev set average fact score: {}'.format(float(dev_scores.mean())))
 
             # Update model parameters
             self.train()
@@ -98,7 +98,7 @@ def run_train(self, train_data, dev_data):
             if self.run_analysis:
                 rewards = None
                 fns = None
-            for example_id in tqdm(range(0, len(train_data), self.batch_size)):
+            for example_id in tqdm(range(0, 127, self.batch_size)):
 
                 self.optim.zero_grad()
 
@@ -154,7 +154,7 @@ def run_train(self, train_data, dev_data):
                     eta = self.action_dropout_anneal_interval
                     if len(dev_metrics_history) > eta and metrics < min(dev_metrics_history[-eta:]):
                         old_action_dropout_rate = self.action_dropout_rate
-                        self.action_dropout_rate *= self.action_dropout_anneal_factor 
+                        self.action_dropout_rate *= self.action_dropout_anneal_factor
                         print('Decreasing action dropout rate: {} -> {}'.format(
                             old_action_dropout_rate, self.action_dropout_rate))
                 # Save checkpoint