update readme

README.md

@@ -22,16 +22,14 @@ DreamSim is a new metric for perceptual image similarity that bridges the gap be
 * DreamSim variants trained for higher resolutions
 * Compatibility with the most recent version of PEFT
 
-## 🚀 Updates
+## 🚀 Newest Updates
+**X/XX/24:** Released new versions of the ensemble and single-branch DreamSim models compatible with `peft>=0.2.0`. 
 
-**7/14/23:** Released three variants of DreamSim that each only use one finetuned ViT model instead of the full ensemble. These single-branch models provide a ~3x speedup over the full ensemble.
-
-
-Here's how they compare to the full ensemble on NIGHTS (2AFC agreement):
-* **Ensemble:** 96.2%
-* **OpenCLIP-ViTB/32:** 95.5%
-* **DINO-ViTB/16:** 94.6%
-* **CLIP-ViTB/32:** 93.9%
+Here's how they perform on the NIGHTS validation set:
+* **Ensemble:** 96.9%
+* **OpenCLIP-ViTB/32:** 95.6%
+* **DINO-ViTB/16:** 95.7%
+* **CLIP-ViTB/32:** 95.0%
 
 ## Table of Contents
 * [Requirements](#requirements)
@@ -96,7 +94,7 @@ distance = model(img1, img2) # The model takes an RGB image from [0, 1], size ba
 
 To run on example images, run `demo.py`. The script should produce distances (0.424, 0.34). 
 
-### (new!) Single-branch models
+### Single-branch models
 By default, DreamSim uses an ensemble of CLIP, DINO, and OpenCLIP (all ViT-B/16). If you need a lighter-weight model you can use *single-branch* versions of DreamSim where only a single backbone is finetuned. The available options are OpenCLIP-ViTB/32, DINO-ViTB/16, CLIP-ViTB/32, in order of performance.  
 
 To load a single-branch model, use the `dreamsim_type` argument. For example:
@@ -153,7 +151,15 @@ DreamSim is trained by fine-tuning on the NIGHTS dataset. For details on the dat
 
 Run `./dataset/download_dataset.sh` to download and unzip the NIGHTS dataset into `./dataset/nights`. The unzipped dataset size is 58 GB. 
 
-**(new!) Visualize NIGHTS and embeddings with the [Voxel51](https://github.com/voxel51/fiftyone) demo:**
+Having trouble with the large file sizes? Run `./dataset/download_chunked_dataset.sh` to download the NIGHTS dataset split into 200 smaller zip files. The output of this script is identical to `download_dataset.sh`.
+
+### (new!) Download the entire 100k pre-filtered NIGHTS dataset
+We only use the 20k unanimous triplets for training and evaluation, but release all 100k triplets (many with few and/or split votes) for research purposes. Run `./dataset/download_unfiltered_dataset.sh` to download and unzip this unfiltered version of the NIGHTS dataset into `./dataset/nights_unfiltered`. The unzipped dataset size is 289 GB.
+
+### (new!) Download the JND data
+Download the just-noticeable difference (JND) votes by running `./dataset/download_jnd_dataset.sh`. The CSV will be downloaded to `./dataset/jnd_votes.csv`. Check out the [Colab](https://colab.research.google.com/drive/1taEOMzFE9g81D9AwH27Uhy2U82tQGAVI?usp=sharing) for an example of loading a JND trial.
+
+### Visualize NIGHTS and embeddings with the [Voxel51](https://github.com/voxel51/fiftyone) demo:
 [![FiftyOne](https://img.shields.io/badge/FiftyOne-blue.svg?logo=data:image/svg+xml;base64,PD94bWwgdmVyc2lvbj0iMS4wIiBlbmNvZGluZz0idXRmLTgiPz4KPCEtLSBHZW5lcmF0b3I6IEFkb2JlIElsbHVzdHJhdG9yIDI3LjMuMSwgU1ZHIEV4cG9ydCBQbHVnLUluIC4gU1ZHIFZlcnNpb246IDYuMDAgQnVpbGQgMCkgIC0tPgo8c3ZnIHZlcnNpb249IjEuMSIgaWQ9IkxheWVyXzEiIHhtbG5zPSJodHRwOi8vd3d3LnczLm9yZy8yMDAwL3N2ZyIgeG1sbnM6eGxpbms9Imh0dHA6Ly93d3cudzMub3JnLzE5OTkveGxpbmsiIHg9IjBweCIgeT0iMHB4IgoJIHZpZXdCb3g9IjAgMCA1MjAuNyA0NzIuNyIgc3R5bGU9ImVuYWJsZS1iYWNrZ3JvdW5kOm5ldyAwIDAgNTIwLjcgNDcyLjc7IiB4bWw6c3BhY2U9InByZXNlcnZlIj4KPHN0eWxlIHR5cGU9InRleHQvY3NzIj4KCS5zdDB7ZmlsbDojRkY2RDAwO30KCS5zdDF7ZmlsbDojOUI5QjlCO30KPC9zdHlsZT4KPGcgaWQ9InN1cmZhY2UxIj4KCTxwYXRoIGNsYXNzPSJzdDAiIGQ9Ik0xMjAuOSw0My4yYzAtMi4yLDEuMy0zLjUsMi4yLTMuOGMwLjYtMC4zLDEuMy0wLjYsMi4yLTAuNmMwLjYsMCwxLjYsMC4zLDIuMiwwLjZsMTMuNyw4TDE2Ny42LDMybC0yNi44LTE1LjMKCQljLTkuNi01LjQtMjEuMS01LjQtMzEsMGMtOS42LDUuOC0xNS4zLDE1LjctMTUuMywyNi44djI4Ni4zbDI2LjIsMTUuM3YtMzAyaDAuMlY0My4yeiIvPgoJPHBhdGggY2xhc3M9InN0MCIgZD0iTTEyNy45LDQyOS42Yy0xLjksMS0zLjgsMC42LTQuNSwwYy0xLTAuNi0yLjItMS42LTIuMi0zLjh2LTE1LjdMOTUsMzk0Ljd2MzFjMCwxMS4yLDUuOCwyMS4xLDE1LjMsMjYuOAoJCWM0LjgsMi45LDEwLjIsNC4yLDE1LjMsNC4yYzUuNCwwLDEwLjUtMS4zLDE1LjMtNC4yTDQwMiwzMDEuN3YtMzAuNEwxMjcuOSw0MjkuNnoiLz4KCTxwYXRoIGNsYXNzPSJzdDAiIGQ9Ik00NzIuNCwyMDcuOGwtMjQ4LTE0My4ybC0yNi41LDE1TDQ1OSwyMzAuNWMxLjksMS4zLDIuMiwyLjksMi4yLDMuOHMtMC4zLDIuOS0yLjIsMy44bC0xMS44LDYuN3YzMC40CgkJbDI0LjktMTQuNGM5LjYtNS40LDE1LjMtMTUuNywxNS4zLTI2LjhDNDg3LjcsMjIzLjEsNDgyLDIxMy4yLDQ3Mi40LDIwNy44eiIvPgoJPHBhdGggY2xhc3M9InN0MSIgZD0iTTc5LjcsMzY4LjVsMjIuNywxMy4xbDI2LjIsMTUuM2w3LjcsNC41bDUuNCwzLjJsOTUuNS01NS4zdi05NS4yYzAtMTIuMSw2LjQtMjMuMywxNi45LTI5LjRsODIuNC00Ny42CgkJTDE5MC4yLDkyLjhsLTIyLjctMTMuMWwyMi43LTEzLjFsMjYuMi0xNS4zbDcuNy00LjVsNy43LDQuNWwxNjEsOTMuM2wzLjItMS45YzkuMy01LjQsMjEuMSwxLjMsMjEuMSwxMi4xdjMuOGwxNSw4LjZWMTQyCgkJYzAtMTIuNS02LjctMjQtMTcuMy0zMEwyNTQuNSwxOS4zYy0xMC45LTYuNC0yNC02LjQtMzQuOCwwTDEzNiw2Ny42djMwMy4ybC0yMi43LTEzLjFMODcsMzQyLjNsLTcuMy00LjJ2LTIzOGwtMjAuMSwxMS41CgkJYy0xMC45LDYuMS0xNy4zLDE3LjYtMTcuMywzMHYxODVjMCwxMi41LDYuNywyNCwxNy4zLDMwTDc5LjcsMzY4LjV6Ii8+Cgk8cGF0aCBjbGFzcz0ic3QxIiBkPSJNNDE3LjEsMjIzLjh2OTQuOWMwLDEyLjEtNi40LDIzLjMtMTYuOSwyOS40bC0xNDEuOSw4Mi4xYy05LjMsNS40LTIxLjEtMS4zLTIxLjEtMTIuMXYtMy44TDE5Ny45LDQzNwoJCWwyMS43LDEyLjVjMTAuOSw2LjQsMjQsNi40LDM0LjgsMEw0MTQuNiwzNTdjMTAuOS02LjQsMTcuMy0xNy42LDE3LjMtMzB2LTk0LjZMNDE3LjEsMjIzLjh6Ii8+CjwvZz4KPC9zdmc+Cg==)](https://try.fiftyone.ai/datasets/nights/samples)
 
 ## Experiments