research
∙
06/07/2023
Git-Theta: A Git Extension for Collaborative Development of Machine Learning Models
Currently, most machine learning models are trained by centralized teams...
share
research
∙
06/06/2023
Soft Merging of Experts with Adaptive Routing
Sparsely activated neural networks with conditional computation learn to...
research
∙
06/02/2023
Resolving Interference When Merging Models
Transfer learning - i.e., further fine-tuning a pre-trained model on a d...
research
∙
05/25/2023
Scaling Data-Constrained Language Models
The current trend of scaling language models involves increasing both pa...
research
∙
02/09/2023
Knowledge is a Region in Weight Space for Fine-tuned Language Models
Research on neural networks has largely focused on understanding a singl...
research
∙
12/02/2022
ColD Fusion: Collaborative Descent for Distributed Multitask Finetuning
Pretraining has been shown to scale well with compute, data size and dat...
research
∙
11/15/2022
Evaluating the Factual Consistency of Large Language Models Through Summarization
While large language models (LLMs) have proven to be effective on a larg...
research
∙
11/15/2022
Large Language Models Struggle to Learn Long-Tail Knowledge
The internet contains a wealth of knowledge – from the birthdays of hist...
research
∙
11/03/2022
Crosslingual Generalization through Multitask Finetuning
Multitask prompted finetuning (MTF) has been shown to help large languag...
research
∙
10/27/2022
What Language Model to Train if You Have One Million GPU Hours?
The crystallization of modeling methods around the Transformer architect...
research
∙
10/02/2022
Compositional Generalization in Unsupervised Compositional Representation Learning: A Study on Disentanglement and Emergent Language
Deep learning models struggle with compositional generalization, i.e. th...
research
∙
10/01/2022
A Combinatorial Perspective on the Optimization of Shallow ReLU Networks
The NP-hard problem of optimizing a shallow ReLU network can be characte...
research
∙
09/29/2022
Bidirectional Language Models Are Also Few-shot Learners
Large language models such as GPT-3 (Brown et al., 2020) can perform arb...
research
∙
09/02/2022
Petals: Collaborative Inference and Fine-tuning of Large Models
Many NLP tasks benefit from using large language models (LLMs) that ofte...
research
∙
08/31/2022
Efficient Methods for Natural Language Processing: A Survey
Getting the most out of limited resources allows advances in natural lan...
research
∙
05/11/2022
Few-Shot Parameter-Efficient Fine-Tuning is Better and Cheaper than In-Context Learning
Few-shot in-context learning (ICL) enables pre-trained language models t...
research
∙
04/12/2022
What Language Model Architecture and Pretraining Objective Work Best for Zero-Shot Generalization?
Large pretrained Transformer language models have been shown to exhibit ...
research
∙
03/31/2022
Scaling Up Models and Data with and
Recent neural network-based language models have benefited greatly from ...
research
∙
02/14/2022
Deduplicating Training Data Mitigates Privacy Risks in Language Models
Past work has shown that large language models are susceptible to privac...
research
∙
02/02/2022
PromptSource: An Integrated Development Environment and Repository for Natural Language Prompts
PromptSource is a system for creating, sharing, and using natural langua...
research
∙
12/20/2021
Between words and characters: A Brief History of Open-Vocabulary Modeling and Tokenization in NLP
What are the units of text that we want to model? From bytes to multi-wo...
research
∙
11/18/2021
Training Neural Networks with Fixed Sparse Masks
During typical gradient-based training of deep neural networks, all of t...
research
∙
11/18/2021
Merging Models with Fisher-Weighted Averaging
Transfer learning provides a way of leveraging knowledge from one task w...
research
∙
10/15/2021
Multitask Prompted Training Enables Zero-Shot Task Generalization
Large language models have recently been shown to attain reasonable zero...
research
∙
06/14/2021
An Empirical Survey of Data Augmentation for Limited Data Learning in NLP
NLP has achieved great progress in the past decade through the use of ne...
research
∙
06/06/2021
On Training Sample Memorization: Lessons from Benchmarking Generative Modeling with a Large-scale Competition
Many recent developments on generative models for natural images have re...
research
∙
05/28/2021
ByT5: Towards a token-free future with pre-trained byte-to-byte models
Most widely-used pre-trained language models operate on sequences of tok...
research
∙
03/22/2021
Improving and Simplifying Pattern Exploiting Training
Recently, pre-trained language models (LMs) have achieved strong perform...
research
∙
02/23/2021
Do Transformer Modifications Transfer Across Implementations and Applications?
The research community has proposed copious modifications to the Transfo...
research
∙
01/01/2021
NeurIPS 2020 EfficientQA Competition: Systems, Analyses and Lessons Learned
We review the EfficientQA competition from NeurIPS 2020. The competition...
research
∙
12/14/2020
Extracting Training Data from Large Language Models
It has become common to publish large (billion parameter) language model...
research
∙
10/22/2020
mT5: A massively multilingual pre-trained text-to-text transformer
The recent "Text-to-Text Transfer Transformer" (T5) leveraged a unified ...
research
∙
04/30/2020
WT5?! Training Text-to-Text Models to Explain their Predictions
Neural networks have recently achieved human-level performance on variou...
research
∙
02/18/2020
Deflecting Adversarial Attacks
There has been an ongoing cycle where stronger defenses against adversar...
research
∙
02/14/2020
Top-K Training of GANs: Improving Generators by Making Critics Less Critical
We introduce a simple (one line of code) modification to the Generative ...
research
∙
02/10/2020
How Much Knowledge Can You Pack Into the Parameters of a Language Model?
It has recently been observed that neural language models trained on uns...
research
∙
01/21/2020
FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence
Semi-supervised learning (SSL) provides an effective means of leveraging...
research
∙
01/10/2020
Towards GAN Benchmarks Which Require Generalization
For many evaluation metrics commonly used as benchmarks for unconditiona...
research
∙
11/21/2019
ReMixMatch: Semi-Supervised Learning with Distribution Alignment and Augmentation Anchoring
We improve the recently-proposed "MixMatch" semi-supervised learning alg...
research
∙
10/23/2019
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Transfer learning, where a model is first pre-trained on a data-rich tas...
research
∙
07/05/2019
Detecting and Diagnosing Adversarial Images with Class-Conditional Capsule Reconstructions
Adversarial examples raise questions about whether neural network models...
research
∙
06/12/2019
Monotonic Infinite Lookback Attention for Simultaneous Machine Translation
Simultaneous machine translation begins to translate each source sentenc...
research
∙
05/06/2019
MixMatch: A Holistic Approach to Semi-Supervised Learning
Semi-supervised learning has proven to be a powerful paradigm for levera...
research
∙
03/22/2019
Imperceptible, Robust, and Targeted Adversarial Examples for Automatic Speech Recognition
Adversarial examples are inputs to machine learning models designed by a...
research
∙
02/21/2019
Lingvo: a Modular and Scalable Framework for Sequence-to-Sequence Modeling
Lingvo is a Tensorflow framework offering a complete solution for collab...
research
∙
07/19/2018
Understanding and Improving Interpolation in Autoencoders via an Adversarial Regularizer
Autoencoders provide a powerful framework for learning compressed repres...
research
∙
06/01/2018
Learning a Latent Space of Multitrack Measures
Discovering and exploring the underlying structure of multi-instrumental...
research
∙
04/24/2018
Realistic Evaluation of Deep Semi-Supervised Learning Algorithms
Semi-supervised learning (SSL) provides a powerful framework for leverag...
research
∙
03/13/2018
A Hierarchical Latent Vector Model for Learning Long-Term Structure in Music
The Variational Autoencoder (VAE) has proven to be an effective model fo...
research
∙
02/23/2018
