fix #786

docs/tutorial/extend/callback.md

@@ -80,8 +80,8 @@ You can overwrite any of the following methods to define a new callback:
 
   The training loops would become `sess.run([training_op, my_op])`.
   This is different from `sess.run(training_op); sess.run(my_op);`,
-  which is what you would get if you run `my_op` in `_trigger_step`.
-	Sometimes the difference matters, please choose carefully.
+  which is what you would get if you write `self.trainer.sess.run(my_op)` in `_trigger_step`.
+  Usually the difference matters, please choose carefully.
 
 * `_trigger_step(self)`
 
@@ -105,6 +105,7 @@ You can overwrite any of the following methods to define a new callback:
 * Access tensors / ops (details mentioned above):
 	* For existing tensors/ops created in the tower, access them through [self.trainer.towers](../../modules/train.html#tensorpack.train.TowerTrainer.towers).
 	* Extra tensors/ops have to be created in `_setup_graph` callback method.
+* Access the current graph and session by `self.trainer.graph` and `self.trainer.sess`.
 * Write stuff to the monitor backend, by `self.trainer.monitors.put_xxx`.
   The monitors might direct your events to TensorFlow events file, JSON file, stdout, etc.
   You can access history monitor data as well. See the docs for [Monitors](../../modules/callbacks.html#tensorpack.callbacks.Monitors)

examples/PennTreebank/PTB-LSTM.py

@@ -57,7 +57,7 @@ class Model(ModelDesc):
         def get_basic_cell():
             cell = rnn.BasicLSTMCell(num_units=HIDDEN_SIZE, forget_bias=0.0, reuse=tf.get_variable_scope().reuse)
             if is_training:
-                cell = rnn.DropoutWrapper(cell, output_keep_prob=DROPOUT)
+                cell = rnn.DropoutWrapper(cell, output_keep_prob=1 - DROPOUT)
             return cell
 
         cell = rnn.MultiRNNCell([get_basic_cell() for _ in range(NUM_LAYER)])
@@ -73,7 +73,7 @@ class Model(ModelDesc):
 
         embeddingW = tf.get_variable('embedding', [VOCAB_SIZE, HIDDEN_SIZE], initializer=initializer)
         input_feature = tf.nn.embedding_lookup(embeddingW, input)  # B x seqlen x hiddensize
-        input_feature = Dropout(input_feature, rate=DROPOUT)
+        input_feature = Dropout(input_feature, keep_prob=1 - DROPOUT)
 
         with tf.variable_scope('LSTM', initializer=initializer):
             input_list = tf.unstack(input_feature, num=SEQ_LEN, axis=1)  # seqlen x (Bxhidden)