Merge pull request #28 from mokemokechicken/feature/enable_tensorboard

add simple tensorboard callback #27

readme.md

@@ -216,6 +216,34 @@ After BestModel moves, numbers are displayed on the board.
 * Top left numbers(1) mean 'Visit Count (=N(s,a))' of the last search.
 * Bottom left numbers(2) mean 'Q Value (=Q(s,a)) on AI side' of the last state and move. The Q values are multiplied by 100.
 
+View Training Log in TensorBoard
+----------------
+
+### 1. install tensorboard
+
+```bash
+pip install tensorboard
+```
+
+### 2. launch tensorboard and access by web browser 
+
+```bash
+tensorboard --logdir logs/tensorboard/
+```
+
+And access `http://<The Machine IP>:6006/`.
+
+### Trouble Shooting
+
+If you can not launch tensorboard by error,
+try to create another new plain project which includes only `tensorflow` and `tensorboard`.
+
+And
+
+```bash
+tensorboard --logdir <PATH TO REVERSI DIR>/logs/tensorboard/
+```
+
 
 Tips and Memo
 ====

src/reversi_zero/config.py

@@ -51,6 +51,7 @@ def __init__(self):
 
         self.log_dir = os.path.join(self.project_dir, "logs")
         self.main_log_path = os.path.join(self.log_dir, "main.log")
+        self.tensorboard_log_dir = os.path.join(self.log_dir, 'tensorboard')
 
     def create_directories(self):
         dirs = [self.project_dir, self.data_dir, self.model_dir, self.play_data_dir, self.log_dir,

src/reversi_zero/configs/normal.py

@@ -50,6 +50,7 @@ def __init__(self):
         self.epoch_to_checkpoint = 1
         self.start_total_steps = 0
         self.save_model_steps = 200
+        self.use_tensorboard = True
 
 
 class ModelConfig:

src/reversi_zero/lib/tensorboard_step_callback.py

@@ -0,0 +1,29 @@
+from keras.callbacks import Callback
+import tensorflow as tf
+
+
+class TensorBoardStepCallback(Callback):
+    """Tensorboard basic visualizations by step.
+
+    """
+
+    def __init__(self, log_dir):
+        super().__init__()
+        self.step = 0
+        self.writer = tf.summary.FileWriter(log_dir)
+
+    def on_batch_end(self, batch, logs=None):
+        self.step += 1
+
+        for name, value in logs.items():
+            if name in ['batch', 'size']:
+                continue
+            summary = tf.Summary()
+            summary_value = summary.value.add()
+            summary_value.simple_value = value.item()
+            summary_value.tag = name
+            self.writer.add_summary(summary, self.step)
+        self.writer.flush()
+
+    def on_train_end(self, logs=None):
+        self.writer.close()

src/reversi_zero/worker/optimize.py

@@ -5,7 +5,7 @@
 
 import keras.backend as K
 import numpy as np
-from keras.callbacks import LambdaCallback, Callback
+from keras.callbacks import Callback
 from keras.optimizers import SGD
 
 from reversi_zero.agent.model import ReversiModel, objective_function_for_policy, \
@@ -16,6 +16,7 @@
 from reversi_zero.lib.data_helper import get_game_data_filenames, read_game_data_from_file, \
     get_next_generation_model_dirs
 from reversi_zero.lib.model_helpler import load_best_model_weight
+from reversi_zero.lib.tensorboard_step_callback import TensorBoardStepCallback
 
 logger = getLogger(__name__)
 
@@ -44,6 +45,11 @@ def training(self):
         save_model_callback = PerStepCallback(self.config.trainer.save_model_steps, self.save_current_model)
         callbacks = [save_model_callback]
 
+        if self.config.trainer.use_tensorboard:
+            callbacks += [TensorBoardStepCallback(
+                log_dir=self.config.resource.tensorboard_log_dir,
+            )]
+
         while True:
             self.load_play_data()
             if self.dataset_size < self.config.trainer.min_data_size_to_learn: