Update version and add CHANGELOG.

CHANGELOG.md

@@ -1,7 +1,22 @@
-ArXiv v1 -> v2 CHANGELOG
-=========================
+CHANGELOG
+=========
 
-[ArXiv v1](https://arxiv.org/abs/2006.06666v1) was our ECCV 2020 submission (reject). [ArXiv v2](https://arxiv.org/abs/2006.06666v2) is out CVPR 2021 submission (accept). The repository snapshots for these two versions are tagged at [`v0.9`](https://github.com/kdexd/virtex/releases/tag/v0.9) and [`v1.0`](https://github.com/kdexd/virtex/releases/tag/v1.0).
+This CHANGELOG file records changes between different arXiv versions of our paper, and the version of this codebase which should be used to reproduce the results in the corresponding arXiv version. View changes between code versions on the [Releases page](https://github.com/kdexd/virtex/releases).
+
+ArXiv v1 -> v2
+==============
+
+**Code version:** `v1.2`.
+
+Fix image captioning results with a modified beam search implementation. _Rest of the downstream task results and pre-trained models are unchanged._
+
+
+ArXiv v1 -> v2
+==============
+
+**Code version:** `v1.0` or `v1.1`.
+
+[ArXiv v1](https://arxiv.org/abs/2006.06666v1) was our ECCV 2020 submission (reject). [ArXiv v2](https://arxiv.org/abs/2006.06666v2) is our CVPR 2021 submission (accept). The repository snapshots for these two versions are tagged at [`v0.9`](https://github.com/kdexd/virtex/releases/tag/v0.9) and [`v1.0`](https://github.com/kdexd/virtex/releases/tag/v1.0).
 
 While the core motivation and approach is the same, we have made some minor changes in our experiments and evaluation setup. These slightly improve model performances across the board (within decimals). New models are available in [`v1.0` model zoo](http://kdexd.github.io/virtex/virtex/usage/model_zoo.html), however links to old models in `v0.9` will be active till June 30, 2021. We encourage you to use the new models!
 

docs/conf.py

@@ -24,7 +24,7 @@
 author = "Karan Desai"
 
 # The full version, including alpha/beta/rc tags
-release = "1.1"
+release = "1.2"
 
 
 # -- General configuration ---------------------------------------------------

scripts/eval_captioning.py

@@ -21,7 +21,7 @@
     evaluate pretrained model on COCO Captions val2017 split."""
 )
 parser.add_argument(
-    "--data-root", default=None,
+    "--images", "--data-root", default=None,
     help="""Path to a directory containing image files to generate captions for.
     Default: COCO val2017 image directory as expected relative to project root."""
 )
@@ -89,6 +89,10 @@ def main(_A: argparse.Namespace):
                 }
             )
 
+    logger.info("Displaying first 25 caption predictions:")
+    for pred in predictions[:25]:
+        logger.info(f"{pred['image_id']} :: {pred['caption']}")
+
     # Save predictions as a JSON file if specified.
     if _A.output is not None:
         os.makedirs(os.path.dirname(_A.output), exist_ok=True)

setup.py

@@ -41,7 +41,7 @@ def get_model_zoo_configs() -> List[str]:
 
 setup(
     name="virtex",
-    version="1.1.0",
+    version="1.2.0",
     author="Karan Desai and Justin Johnson",
     description="VirTex: Learning Visual Representations with Textual Annotations",
     package_data={"virtex.model_zoo": get_model_zoo_configs()},

virtex/models/captioning.py

@@ -72,7 +72,7 @@ def __init__(
         # These boundary indices are needed for beam search.
         self.sos_index = sos_index
         self.eos_index = eos_index
-        self.beam_search = decoder
+        self.decoder = decoder
         self.loss = nn.CrossEntropyLoss(ignore_index=self.padding_idx)
 
     def forward(self, batch: Dict[str, torch.Tensor]) -> Dict[str, Any]: