浏览代码

Change the train and validation shards to the correct amounts

Neal Wu 8 年之前
父节点
当前提交
ea97e5c3a7
共有 1 个文件被更改,包括 9 次插入9 次删除
  1. 9 9
      inception/README.md

+ 9 - 9
inception/README.md

@@ -668,8 +668,8 @@ bazel-bin/inception/build_image_data \
   --validation_directory="${VALIDATION_DIR}" \
   --output_directory="${OUTPUT_DIRECTORY}" \
   --labels_file="${LABELS_FILE}" \
-  --train_shards=24 \
-  --validation_shards=8 \
+  --train_shards=128 \
+  --validation_shards=24 \
   --num_threads=8
 ```
 
@@ -694,20 +694,20 @@ class.
 After running this script produces files that look like the following:
 
 ```shell
-  $TRAIN_DIR/train-00000-of-00024
-  $TRAIN_DIR/train-00001-of-00024
+  $TRAIN_DIR/train-00000-of-00128
+  $TRAIN_DIR/train-00001-of-00128
   ...
-  $TRAIN_DIR/train-00023-of-00024
+  $TRAIN_DIR/train-00127-of-00128
 
 and
 
-  $VALIDATION_DIR/validation-00000-of-00008
-  $VALIDATION_DIR/validation-00001-of-00008
+  $VALIDATION_DIR/validation-00000-of-00024
+  $VALIDATION_DIR/validation-00001-of-00024
   ...
-  $VALIDATION_DIR/validation-00007-of-00008
+  $VALIDATION_DIR/validation-00023-of-00024
 ```
 
-where 24 and 8 are the number of shards specified for each dataset,
+where 128 and 24 are the number of shards specified for each dataset,
 respectively. Generally speaking, we aim for selecting the number of shards such
 that roughly 1024 images reside in each shard. Once this data set is built, you
 are ready to train or fine-tune an Inception model on this data set.