Merge remote-tracking branch 'github/master'

Author: pesser

README.md

@@ -13,7 +13,8 @@
 [arXiv](https://arxiv.org/abs/2012.09841) | [BibTeX](#bibtex) | [Project Page](https://compvis.github.io/taming-transformers/)
 
 ### News
-
+- We added a [colab notebook](https://colab.research.google.com/github/CompVis/taming-transformers/blob/master/scripts/reconstruction_usage.ipynb) which compares two VQGANs and OpenAI's [DALL-E](). See also [this section](#more-resources).
+- We now include an overview of pretrained models in [Tab.1](#overview-of-pretrained-models)
 - The streamlit demo now supports image completions.
 - We now include a couple of examples from the D-RIN dataset so you can run the
   [D-RIN demo](#d-rin) without preparing the dataset first.
@@ -39,8 +40,8 @@ for a comparison and discussion of reconstruction capabilities.
 | ------------- | ------------- |-------------  | -------------  |-------------  |
 | FFHQ (f=16) | 11.4 | coming soon... | 
 | CelebA-HQ (f=16) | 10.7 | coming soon... | 
-| ADE20K (f=16) | 35.5  | [ade20k_transformer](https://k00.fr/ot46cksa) | [ade20k_samples.zip](https://heibox.uni-heidelberg.de/f/70bb78cbaf844501b8fb/) [2k]
-| COCO-Stuff (f=16) | 20.4  | [coco_transformer](https://k00.fr/2zz6i2ce) | [coco_samples.zip](https://heibox.uni-heidelberg.de/f/a395a9be612f4a7a8054/) [5k]
+| ADE20K (f=16) | 35.5  | [ade20k_transformer](https://k00.fr/ot46cksa) | [ade20k_samples.zip](https://heibox.uni-heidelberg.de/f/70bb78cbaf844501b8fb/) [2k] | evaluated on val split (2k images)
+| COCO-Stuff (f=16) | 20.4  | [coco_transformer](https://k00.fr/2zz6i2ce) | [coco_samples.zip](https://heibox.uni-heidelberg.de/f/a395a9be612f4a7a8054/) [5k] | evaluated on val split (5k images)
 | ImageNet (cIN) (f=16) |  | coming soon...
 | |  | | || |
 | FacesHQ (f=16) | -- | [faceshq_transformer](https://k00.fr/qqfl2do8)

scripts/reconstruction_usage.ipynb

@@ -273,7 +273,7 @@
     "# Taming Transformers\n",
     "## Reconstruction Capabilities of VQGAN (Colab Notebook)\n",
     "\n",
-    "This notebook provides code to (visually) analyze the first stage models used to generate images as in [Taming Transformers for High-Resolution Image Synthesis](https://github.com/CompVis/taming-transformers) \n",
+    "This notebook provides code to (visually) analyze the first stage models used to generate images as in [Taming Transformers for High-Resolution Image Synthesis](https://github.com/CompVis/taming-transformers)\n",
     "and a comparison to the first stage model used in [DALL-E](https://openai.com/blog/dall-e/)."
    ]
   },
@@ -284,8 +284,8 @@
    },
    "source": [
     "### Setup\n",
-    "Clone the repository and download pretrained VQGANs: a small one with a codebook dimensionality $\\dim \\mathcal{Z} = 1024$ \n",
-    "and a larger one with $ \\dim \\mathcal{Z} = 16384$. Both perform *four* downsampling steps, e.g. an input image of \n",
+    "Clone the repository and download pretrained VQGANs: a small one with a codebook dimensionality $\\dim \\mathcal{Z} = 1024$\n",
+    "and a larger one with $ \\dim \\mathcal{Z} = 16384$. Both perform *four* downsampling steps, e.g. an input image of\n",
     "size $256 \\times 256$ will be mapped to a latent code of size $16 \\times 16$."
    ]
   },