update multiannotator links

Author: jwmueller

multiannotator_cifar10/multiannotator_cifar10.ipynb

@@ -13,7 +13,7 @@
    "id": "44a0555b",
    "metadata": {},
    "source": [
-    "This example shows how to repeatedly improve consensus labels established from data labeled by multiple annotators by iterating the following steps: (1) train a model on the current consensus labels, (2) leverage the model's predictions to obtain superior consensus labels that can be used to subsequently train a better model in the next round. In each round, consensus labels are established using the [CROWDLAB algorithm](TODO:linktopaper), for which a quickstart tutorial is available in the [cleanlab documentation](https://docs.cleanlab.ai/). \n",
+    "This example shows how to repeatedly improve consensus labels established from data labeled by multiple annotators by iterating the following steps: (1) train a model on the current consensus labels, (2) leverage the model's predictions to obtain superior consensus labels that can be used to subsequently train a better model in the next round. In each round, consensus labels are established using the [CROWDLAB algorithm](https://cleanlab.github.io/multiannotator-benchmarks/paper.pdf), for which a quickstart tutorial is available in the [cleanlab documentation](https://docs.cleanlab.ai/stable/tutorials/multiannotator.html). \n",
     "\n",
     "Here we demonstrate this functionality using a subset of the [CIFAR-10H](https://github.com/jcpeterson/cifar-10h) dataset from Peterson et al. (2019), in which multiple human annotators were asked to suggest labels for images from the famous [CIFAR-10](https://www.cs.toronto.edu/~kriz/cifar.html) image classification dataset.\n",
     "Because this notebook utilizes AutoML for model training and cleanlab is compatible with any model/dataset, you should be able to run the below code with any image classification dataset where each image has been labeled by multiple annotators."