Accept a set of test captions instead of only one.

ruotianluo · Sep 4, 2017 · 046d8cf · 046d8cf
1 parent ac6d9ce
commit 046d8cf
Show file tree

Hide file tree

Showing 4 changed files with 1,010 additions and 1,005 deletions.
diff --git a/README.md b/README.md
@@ -34,7 +34,7 @@ Running SPICE with no arguments prints the following help message:
 
     See README file for additional information and input format details
 
-The input.json file should contain of an array of json objects, each representing a single caption and containing `image_id`, `test` and `refs` fields. See `example_input.json`
+The input.json file should contain of an array of json objects, each representing a single caption and containing `image_id`, `tests` and `refs` fields. See `example_input.json`
 
 It is recommended to provide a path to an empty directory in the `-cache` argument as it makes repeated evaluations much faster.
 

diff --git a/example_input.json b/example_input.json
@@ -1,7 +1,7 @@
 [
   {
     "image_id": 1,
-    "test": "a kitchen and dining room and living room.",
+    "tests": ["a kitchen and dining room and living room."],
     "refs": [
       "the room is empty other than the furniture. ",
       "an open floor plan displays a modern kitchen, dining and living room arrangement.",
@@ -11,7 +11,7 @@
   },
   {
     "image_id": 234,
-    "test": "two fedex trucks parked on the side of the street . ",
+    "tests": ["two fedex trucks parked on the side of the street . "],
     "refs": [
       "a couple of fed ex delivery trucks parked outside of a building.",
       "two fedex trucks parked on a side of a street with tall buidings behind them.",