Collection of algorithms to learn loss and reward functions via gradient-based bi-level optimization.
Toy datasets to evaluate algorithms for domain generalization and invariance learning.
The Python library with command line tools to interact with Dynabench(https://dynabench.org/), such as uploading models.
We provide a PyTorch implementation of the paper Voice Separation with an Unknown Number of Multiple Speakers In which, we present a new method for separating a mixed audio sequence, in which multiple voices speak simultaneously. The new method employs gated neural networks that are trained to separate the voices at multiple processing steps, while maintaining the speaker in each output channel fixed. A different model is trained for every number of possible speakers, and the model with the largest number of speakers is employed to select the actual number of speakers in a given sample. Our method greatly outperforms the current state of the art, which, as we show, is not competitive for more than two speakers.
This is code associated with the paper, "Secure multiparty computations in floating-point arithmetic," published in the Institute of Mathematics and its Applications' "Information and Inference: a Journal of the IMA."
Code repo for the ICML 2021 paper "Making Paper Reviewing Robust to Bid Manipulation Attacks".
Reinforcement learning environments for compiler and program optimization tasks
This repo covers the implementation for Labelling unlabelled videos from scratch with multi-modal self-supervision, which learns clusters from multi-modal data in a self-supervised way.
Multi-hop dense retrieval for question answering
Data & Code for ACCENTOR: "Adding Chit-Chat to Enhance Task-Oriented Dialogues" (NAACL 2021)
Ridge SfM Structure from Motion via robust pairwise matching under depth uncertainty
Code for the benchmark containing dataset, models and metrics for productive concept learning -- a kind of compositional reasoning task that requires reasoning about uncertainty and learning compositionally rich and challenging concepts in a low-shot, meta-learning framework.
When told to understand the shape of a new object, the most instinctual approach is to pick it up and inspect it with your hand and eyes in tandem. Here, touch provides high fidelity localized information while vision provides complementary global context. However, in 3D shape reconstruction, the complementary fusion of visual and haptic modalities remains largely unexplored. In this paper, we study this problem and present an effective chart-based approach to fusing vision and touch, which leverages advances in graph convolutional networks. To do so, we introduce a dataset of simulated touch and vision signals from the interaction between a robotic hand and a large array of 3D objects. Our results show that (1) leveraging both vision and touch signals consistently improves single-modality baselines, especially when the object is occluded by the hand touching it; (2) our approach outperforms alternative modality fusion methods and strongly benefits from the proposed chart-based structure; (3) reconstruction quality boosts with the number of grasps provided; and (4) the touch information not only enhances the reconstruction at the touch site but also extrapolates to its local neighborhood.
SentAugment is a data augmentation technique for NLP that retrieves similar sentences from a large bank of sentences. It can be used in combination with self-training and knowledge-distillation, or for retrieving paraphrases.
We evaluate many models used for biomedical and clinical nlp tasks, and train new models that perform much better.
An open-source implementation of SEAL for link prediction in open graph benchmark (OGB) datasets.
