Fine-Tuning can Distort Pretrained Features and Underperform Out-of-Distribution

When transferring a pretrained model to a downstream task, two popular methods are full fine-tuning (updating all the model parameters) and linear probing (updating only the last linear layer -- the "head"). It is well known that fine-tuning leads to better accuracy in-distribution (ID). However, in this paper, we find that fine-tuning can achieve worse accuracy than linear probing out-of-distribution (OOD) when the pretrained features are good and the distribution shift is large. On 10 distribution shift datasets (Breeds-Living17, Breeds-Entity30, DomainNet, CIFAR $\to$ STL, CIFAR10.1, FMoW, ImageNetV2, ImageNet-R, ImageNet-A, ImageNet-Sketch), fine-tuning obtains on average 2% higher accuracy ID but 7% lower accuracy OOD than linear probing. We show theoretically that this tradeoff between ID and OOD accuracy arises even in a simple setting: fine-tuning overparameterized two-layer linear networks. We prove that the OOD error of fine-tuning is high when we initialize with a fixed or random head -- this is because while fine-tuning learns the head, the lower layers of the neural network change simultaneously and distort the pretrained features. Our analysis suggests that the easy two-step strategy of linear probing then full fine-tuning (LP-FT), sometimes used as a fine-tuning heuristic, combines the benefits of both fine-tuning and linear probing. Empirically, LP-FT outperforms both fine-tuning and linear probing on the above datasets (1% better ID, 10% better OOD than full fine-tuning).

An Explanation of In-context Learning as Implicit Bayesian Inference

Large pretrained language models such as GPT-3 have the surprising ability to do in-context learning, where the model learns to do a downstream task simply by conditioning on a prompt consisting of input-output examples. Without being explicitly pretrained to do so, the language model learns from these examples during its forward pass without parameter updates on "out-of-distribution" prompts. Thus, it is unclear what mechanism enables in-context learning. In this paper, we study the role of the pretraining distribution on the emergence of in-context learning under a mathematical setting where the pretraining texts have long-range coherence. Here, language model pretraining requires inferring a latent document-level concept from the conditioning text to generate coherent next tokens. At test time, this mechanism enables in-context learning by inferring the shared latent concept between prompt examples and applying it to make a prediction on the test example. Concretely, we prove that in-context learning occurs implicitly via Bayesian inference of the latent concept when the pretraining distribution is a mixture of HMMs. This can occur despite the distribution mismatch between prompts and pretraining data. In contrast to messy large-scale pretraining datasets for in-context learning in natural language, we generate a family of small-scale synthetic datasets (GINC) where Transformer and LSTM language models both exhibit in-context learning. Beyond the theory which focuses on the effect of the pretraining distribution, we empirically find that scaling model size improves in-context accuracy even when the pretraining loss is the same.

On the Opportunities and Risks of Foundation Models

AI is undergoing a paradigm shift with the rise of models (e.g., BERT, DALL-E, GPT-3) that are trained on broad data at scale and are adaptable to a wide range of downstream tasks. We call these models foundation models to underscore their critically central yet incomplete character. This report provides a thorough account of the opportunities and risks of foundation models, ranging from their capabilities (e.g., language, vision, robotics, reasoning, human interaction) and technical principles(e.g., model architectures, training procedures, data, systems, security, evaluation, theory) to their applications (e.g., law, healthcare, education) and societal impact (e.g., inequity, misuse, economic and environmental impact, legal and ethical considerations). Though foundation models are based on standard deep learning and transfer learning, their scale results in new emergent capabilities,and their effectiveness across so many tasks incentivizes homogenization. Homogenization provides powerful leverage but demands caution, as the defects of the foundation model are inherited by all the adapted models downstream. Despite the impending widespread deployment of foundation models, we currently lack a clear understanding of how they work, when they fail, and what they are even capable of due to their emergent properties. To tackle these questions, we believe much of the critical research on foundation models will require deep interdisciplinary collaboration commensurate with their fundamentally sociotechnical nature.

Just Train Twice: Improving Group Robustness without Training Group Information

Standard training via empirical risk minimization (ERM) can produce models that achieve high accuracy on average but low accuracy on certain groups, especially in the presence of spurious correlations between the input and label. Prior approaches that achieve high worst-group accuracy, like group distributionally robust optimization (group DRO) require expensive group annotations for each training point, whereas approaches that do not use such group annotations typically achieve unsatisfactory worst-group accuracy. In this paper, we propose a simple two-stage approach, JTT, that first trains a standard ERM model for several epochs, and then trains a second model that upweights the training examples that the first model misclassified. Intuitively, this upweights examples from groups on which standard ERM models perform poorly, leading to improved worst-group performance. Averaged over four image classification and natural language processing tasks with spurious correlations, JTT closes 75% of the gap in worst-group accuracy between standard ERM and group DRO, while only requiring group annotations on a small validation set in order to tune hyperparameters.

Accuracy on the Line: On the Strong Correlation Between Out-of-Distribution and In-Distribution Generalization

For machine learning systems to be reliable, we must understand their performance in unseen, out-of-distribution environments. In this paper, we empirically show that out-of-distribution performance is strongly correlated with in-distribution performance for a wide range of models and distribution shifts. Specifically, we demonstrate strong correlations between in-distribution and out-of-distribution performance on variants of CIFAR-10 & ImageNet, a synthetic pose estimation task derived from YCB objects, satellite imagery classification in FMoW-WILDS, and wildlife classification in iWildCam-WILDS. The strong correlations hold across model architectures, hyperparameters, training set size, and training duration, and are more precise than what is expected from existing domain adaptation theory. To complete the picture, we also investigate cases where the correlation is weaker, for instance some synthetic distribution shifts from CIFAR-10-C and the tissue classification dataset Camelyon17-WILDS. Finally, we provide a candidate theory based on a Gaussian data model that shows how changes in the data covariance arising from distribution shift can affect the observed correlations.

Enabling certification of verification-agnostic networks via memory-efficient semidefinite programming

Convex relaxations have emerged as a promising approach for verifying desirable properties of neural networks like robustness to adversarial perturbations. Widely used Linear Programming (LP) relaxations only work well when networks are trained to facilitate verification. This precludes applications that involve verification-agnostic networks, i.e., networks not specially trained for verification. On the other hand, semidefinite programming (SDP) relaxations have successfully be applied to verification-agnostic networks, but do not currently scale beyond small networks due to poor time and space asymptotics. In this work, we propose a first-order dual SDP algorithm that (1) requires memory only linear in the total number of network activations, (2) only requires a fixed number of forward/backward passes through the network per iteration. By exploiting iterative eigenvector methods, we express all solver operations in terms of forward and backward passes through the network, enabling efficient use of hardware like GPUs/TPUs. For two verification-agnostic networks on MNIST and CIFAR-10, we significantly improve L-inf verified robust accuracy from 1% to 88% and 6% to 40% respectively. We also demonstrate tight verification of a quadratic stability specification for the decoder of a variational autoencoder.

Explore then Execute: Adapting without Rewards via Factorized Meta-Reinforcement Learning

We seek to efficiently learn by leveraging shared structure between different tasks and environments. For example, cooking is similar in different kitchens, even though the ingredients may change location. In principle, meta-reinforcement learning approaches can exploit this shared structure, but in practice, they fail to adapt to new environments when adaptation requires targeted exploration (e.g., exploring the cabinets to find ingredients in a new kitchen). We show that existing approaches fail due to a chicken-and-egg problem: learning what to explore requires knowing what information is critical for solving the task, but learning to solve the task requires already gathering this information via exploration. For example, exploring to find the ingredients only helps a robot prepare a meal if it already knows how to cook, but the robot can only learn to cook if it already knows where the ingredients are. To address this, we propose a new exploration objective (DREAM), based on identifying key information in the environment, independent of how this information will exactly be used solve the task. By decoupling exploration from task execution, DREAM explores and consequently adapts to new environments, requiring no reward signal when the task is specified via an instruction. Empirically, DREAM scales to more complex problems, such as sparse-reward 3D visual navigation, while existing approaches fail from insufficient exploration.