Alexei A. Efros

ICLR Conference 2025 Conference Paper

Interpreting the Second-Order Effects of Neurons in CLIP

• Yossi Gandelsman
• Alexei A. Efros
• Jacob Steinhardt

We interpret the function of individual neurons in CLIP by automatically describing them using text. Analyzing the direct effects (i.e. the flow from a neuron through the residual stream to the output) or the indirect effects (overall contribution) fails to capture the neurons' function in CLIP. Therefore, we present the "second-order lens", analyzing the effect flowing from a neuron through the later attention heads, directly to the output. We find that these effects are highly selective: for each neuron, the effect is significant for <2% of the images. Moreover, each effect can be approximated by a single direction in the text-image space of CLIP. We describe neurons by decomposing these directions into sparse sets of text representations. The sets reveal polysemantic behavior - each neuron corresponds to multiple, often unrelated, concepts (e.g. ships and cars). Exploiting this neuron polysemy, we mass-produce "semantic" adversarial examples by generating images with concepts spuriously correlated to the incorrect class. Additionally, we use the second-order effects for zero-shot segmentation, outperforming previous methods. Our results indicate that a automated interpretation of neurons can be used for model deception and for introducing new model capabilities

ICML Conference 2025 Conference Paper

IT3: Idempotent Test-Time Training

• Nikita Durasov
• Assaf Shocher
• Doruk Öner
• Gal Chechik
• Alexei A. Efros
• Pascal Fua

Deep learning models often struggle when deployed in real-world settings due to distribution shifts between training and test data. While existing approaches like domain adaptation and test-time training (TTT) offer partial solutions, they typically require additional data or domain-specific auxiliary tasks. We present Idempotent Test-Time Training (IT3), a novel approach that enables on-the-fly adaptation to distribution shifts using only the current test instance, without any auxiliary task design. Our key insight is that enforcing idempotence—where repeated applications of a function yield the same result—can effectively replace domain-specific auxiliary tasks used in previous TTT methods. We theoretically connect idempotence to prediction confidence and demonstrate that minimizing the distance between successive applications of our model during inference leads to improved out-of-distribution performance. Extensive experiments across diverse domains (including image classification, aerodynamics prediction, and aerial segmentation) and architectures (MLPs, CNNs, GNNs) show that IT3 consistently outperforms existing approaches while being simpler and more widely applicable. Our results suggest that idempotence provides a universal principle for test-time adaptation that generalizes across domains and architectures.

ICLR Conference 2025 Conference Paper

Prioritized Generative Replay

• Renhao Wang
• Kevin Frans
• Pieter Abbeel
• Sergey Levine
• Alexei A. Efros

Sample-efficient online reinforcement learning often uses replay buffers to store experience for reuse when updating the value function. However, uniform replay is inefficient, since certain classes of transitions can be more relevant to learning. While prioritization of more useful samples is helpful, this strategy can also lead to overfitting, as useful samples are likely to be more rare. In this work, we instead propose a prioritized, parametric version of an agent's memory, using generative models to capture online experience. This paradigm enables (1) densification of past experience, with new generations that benefit from the generative model's generalization capacity and (2) guidance via a family of "relevance functions" that push these generations towards more useful parts of an agent's acquired history. We show this recipe can be instantiated using conditional diffusion models and simple relevance functions such as curiosity- or value-based metrics. Our approach consistently improves performance and sample efficiency in both state- and pixel-based domains. We expose the mechanisms underlying these gains, showing how guidance promotes diversity in our generated transitions and reduces overfitting. We also showcase how our approach can train policies with even higher update-to-data ratios than before, opening up avenues to better scale online RL agents. Project page available at: https://pgenreplay.github.io

JMLR Journal 2025 Journal Article

Test-Time Training on Video Streams

• Renhao Wang
• Yu Sun
• Arnuv Tandon
• Yossi Gandelsman
• Xinlei Chen
• Alexei A. Efros
• Xiaolong Wang

Prior work has established Test-Time Training (TTT) as a general framework to further improve a trained model at test time. Before making a prediction on each test instance, the model is first trained on the same instance using a self-supervised task such as reconstruction. We extend TTT to the streaming setting, where multiple test instances - video frames in our case - arrive in temporal order. Our extension is online TTT: The current model is initialized from the previous model, then trained on the current frame and a small window of frames immediately before. Online TTT significantly outperforms the fixed-model baseline for four tasks, on three real-world datasets. The improvements are more than 2.2x and 1.5x for instance and panoptic segmentation. Surprisingly, online TTT also outperforms its offline variant that accesses strictly more information, training on all frames from the entire test video regardless of temporal order. This finding challenges those in prior work using synthetic videos. We formalize a notion of locality as the advantage of online over offline TTT, and analyze its role with ablations and a theory based on bias-variance trade-off. [abs] [ pdf ][ bib ] [ code ] &copy JMLR 2025. ( edit, beta )

NeurIPS Conference 2024 Conference Paper

Data Attribution for Text-to-Image Models by Unlearning Synthesized Images

• Sheng-Yu Wang
• Aaron Hertzmann
• Alexei A. Efros
• Jun-Yan Zhu
• Richard Zhang

The goal of data attribution for text-to-image models is to identify the training images that most influence the generation of a new image. Influence is defined such that, for a given output, if a model is retrained from scratch without the most influential images, the model would fail to reproduce the same output. Unfortunately, directly searching for these influential images is computationally infeasible, since it would require repeatedly retraining models from scratch. In our work, we propose an efficient data attribution method by simulating unlearning the synthesized image. We achieve this by increasing the training loss on the output image, without catastrophic forgetting of other, unrelated concepts. We then identify training images with significant loss deviations after the unlearning process and label these as influential. We evaluate our method with a computationally intensive but "gold-standard" retraining from scratch and demonstrate our method's advantages over previous methods.

ICML Conference 2024 Conference Paper

Disentangled 3D Scene Generation with Layout Learning

• Dave Epstein
• Ben Poole
• Ben Mildenhall
• Alexei A. Efros
• Aleksander Holynski

We introduce a method to generate 3D scenes that are disentangled into their component objects. This disentanglement is unsupervised, relying only on the knowledge of a large pretrained text-to-image model. Our key insight is that objects can be discovered by finding parts of a 3D scene that, when rearranged spatially, still produce valid configurations of the same scene. Concretely, our method jointly optimizes multiple NeRFs—each representing its own object—along with a set of layouts that composite these objects into scenes. We then encourage these composited scenes to be in-distribution according to the image generator. We show that despite its simplicity, our approach successfully generates 3D scenes decomposed into individual objects, enabling new capabilities in text-to-3D content creation.

NeurIPS Conference 2024 Conference Paper

Evaluating Multiview Object Consistency in Humans and Image Models

• Tyler Bonnen
• Stephanie Fu
• Yutong Bai
• Thomas O'Connell
• Yoni Friedman
• Nancy Kanwisher
• Joshua B. Tenenbaum
• Alexei A. Efros

We introduce a benchmark to directly evaluate the alignment between human observers and vision models on a 3D shape inference task. We leverage an experimental design from the cognitive sciences: given a set of images, participants identify which contain the same/different objects, despite considerable viewpoint variation. We draw from a diverse range of images that include common objects (e. g. , chairs) as well as abstract shapes (i. e. , procedurally generated 'nonsense' objects). After constructing over 2000 unique image sets, we administer these tasks to human participants, collecting 35K trials of behavioral data from over 500 participants. This includes explicit choice behaviors as well as intermediate measures, such as reaction time and gaze data. We then evaluate the performance of common vision models (e. g. , DINOv2, MAE, CLIP). We find that humans outperform all models by a wide margin. Using a multi-scale evaluation approach, we identify underlying similarities and differences between models and humans: while human-model performance is correlated, humans allocate more time/processing on challenging trials. All images, data, and code can be accessed via our project page.

ICLR Conference 2024 Conference Paper

Idempotent Generative Network

• Assaf Shocher
• Amil Dravid
• Yossi Gandelsman
• Inbar Mosseri
• Michael Rubinstein
• Alexei A. Efros

We propose a new approach for generative modeling based on training a neural network to be idempotent. An idempotent operator is one that can be applied sequentially without changing the result beyond the initial application, namely $f(f(z))=f(z)$. The proposed model $f$ is trained to map a source distribution (e.g, Gaussian noise) to a target distribution (e.g. realistic images) using the following objectives: (1) Instances from the target distribution should map to themselves, namely $f(x)=x$. We define the target manifold as the set of all instances that $f$ maps to themselves. (2) Instances that form the source distribution should map onto the defined target manifold. This is achieved by optimizing the idempotence term, $f(f(z))=f(z)$ which encourages the range of $f(z)$ to be on the target manifold. Under ideal assumptions such a process provably converges to the target distribution. This strategy results in a model capable of generating an output in one step, maintaining a consistent latent space, while also allowing sequential applications for refinement. Additionally, we find that by processing inputs from both target and source distributions, the model adeptly projects corrupted or modified data back to the target manifold. This work is a first step towards a ``global projector'' that enables projecting any input into a target data distribution.

ICLR Conference 2024 Conference Paper

Interpreting CLIP's Image Representation via Text-Based Decomposition

• Yossi Gandelsman
• Alexei A. Efros
• Jacob Steinhardt

We investigate the CLIP image encoder by analyzing how individual model components affect the final representation. We decompose the image representation as a sum across individual image patches, model layers, and attention heads, and use CLIP's text representation to interpret the summands. Interpreting the attention heads, we characterize each head's role by automatically finding text representations that span its output space, which reveals property-specific roles for many heads (e.g. location or shape). Next, interpreting the image patches, we uncover an emergent spatial localization within CLIP. Finally, we use this understanding to remove spurious features from CLIP and to create a strong zero-shot image segmenter. Our results indicate that scalable understanding of transformer models is attainable and can be used to repair and improve models.

NeurIPS Conference 2024 Conference Paper

Interpreting the Weight Space of Customized Diffusion Models

• Amil Dravid
• Yossi Gandelsman
• Kuan-Chieh Wang
• Rameen Abdal
• Gordon Wetzstein
• Alexei A. Efros
• Kfir Aberman

We investigate the space of weights spanned by a large collection of customized diffusion models. We populate this space by creating a dataset of over 60, 000 models, each of which is a base model fine-tuned to insert a different person's visual identity. We model the underlying manifold of these weights as a subspace, which we term $\textit{weights2weights}$. We demonstrate three immediate applications of this space that result in new diffusion models -- sampling, editing, and inversion. First, sampling a set of weights from this space results in a new model encoding a novel identity. Next, we find linear directions in this space corresponding to semantic edits of the identity (e. g. , adding a beard), resulting in a new model with the original identity edited. Finally, we show that inverting a single image into this space encodes a realistic identity into a model, even if the input image is out of distribution (e. g. , a painting). We further find that these linear properties of the diffusion model weight space extend to other visual concepts. Our results indicate that the weight space of fine-tuned diffusion models can behave as an interpretable $\textit{meta}$-latent space producing new models.

NeurIPS Conference 2024 Conference Paper

Rethinking Score Distillation as a Bridge Between Image Distributions

• David McAllister
• Songwei Ge
• Jia-Bin Huang
• David W. Jacobs
• Alexei A. Efros
• Aleksander Holynski
• Angjoo Kanazawa

Score distillation sampling (SDS) has proven to be an important tool, enabling the use of large-scale diffusion priors for tasks operating in data-poor domains. Unfortunately, SDS has a number of characteristic artifacts that limit its utility in general-purpose applications. In this paper, we make progress toward understanding the behavior of SDS and its variants by viewing them as solving an optimal-cost transport path from some current source distribution to a target distribution. Under this new interpretation, we argue that these methods' characteristic artifacts are caused by (1) linear approximation of the optimal path and (2) poor estimates of the source distribution. We show that by calibrating the text conditioning of the source distribution, we can produce high-quality generation and translation results with little extra overhead. Our method can be easily applied across many domains, matching or beating the performance of specialized methods. We demonstrate its utility in text-to-2D, text-to-3D, translating paintings to real images, optical illusion generation, and 3D sketch-to-real. We compare our method to existing approaches for score distillation sampling and show that it can produce high-frequency details with realistic colors.

ICML Conference 2023 Conference Paper

Internet Explorer: Targeted Representation Learning on the Open Web

• Alexander Cong Li
• Ellis Langham Brown
• Alexei A. Efros
• Deepak Pathak

Vision models typically rely on fine-tuning general-purpose models pre-trained on large, static datasets. These general-purpose models only capture the knowledge within their pre-training datasets, which are tiny, out-of-date snapshots of the Internet—where billions of images are uploaded each day. We suggest an alternate approach: rather than hoping our static datasets transfer to our desired tasks after large-scale pre-training, we propose dynamically utilizing the Internet to quickly train a small-scale model that does extremely well on a target dataset. Our approach, called Internet Explorer, explores the web in a self-supervised manner to progressively find relevant examples that improve performance on a desired target dataset. It cycles between searching for images on the Internet with text queries, self-supervised training on downloaded images, determining which images were useful, and prioritizing what to search for next. We evaluate Internet Explorer across several datasets and show that it outperforms or matches CLIP oracle performance using just a single GPU desktop to actively query the Internet for 30-40 hours.

ICLR Conference 2021 Conference Paper

Learning Cross-Domain Correspondence for Control with Dynamics Cycle-Consistency

• Qiang Zhang
• Tete Xiao
• Alexei A. Efros
• Lerrel Pinto
• Xiaolong Wang 0004

At the heart of many robotics problems is the challenge of learning correspondences across domains. For instance, imitation learning requires obtaining correspondence between humans and robots; sim-to-real requires correspondence between physics simulators and real hardware; transfer learning requires correspondences between different robot environments. In this paper, we propose to learn correspondence across such domains emphasizing on differing modalities (vision and internal state), physics parameters (mass and friction), and morphologies (number of limbs). Importantly, correspondences are learned using unpaired and randomly collected data from the two domains. We propose dynamics cycles that align dynamic robotic behavior across two domains using a cycle consistency constraint. Once this correspondence is found, we can directly transfer the policy trained on one domain to the other, without needing any additional fine-tuning on the second domain. We perform experiments across a variety of problem domains, both in simulation and on real robots. Our framework is able to align uncalibrated monocular video of a real robot arm to dynamic state-action trajectories of a simulated arm without paired data. Video demonstrations of our results are available at: https://sites.google.com/view/cycledynamics .

ICLR Conference 2021 Conference Paper

Self-Supervised Policy Adaptation during Deployment

• Nicklas Hansen 0001
• Rishabh Jangir
• Yu Sun 0020
• Guillem Alenyà
• Pieter Abbeel
• Alexei A. Efros
• Lerrel Pinto
• Xiaolong Wang 0004

In most real world scenarios, a policy trained by reinforcement learning in one environment needs to be deployed in another, potentially quite different environment. However, generalization across different environments is known to be hard. A natural solution would be to keep training after deployment in the new environment, but this cannot be done if the new environment offers no reward signal. Our work explores the use of self-supervision to allow the policy to continue training after deployment without using any rewards. While previous methods explicitly anticipate changes in the new environment, we assume no prior knowledge of those changes yet still obtain significant improvements. Empirical evaluations are performed on diverse simulation environments from DeepMind Control suite and ViZDoom, as well as real robotic manipulation tasks in continuously changing environments, taking observations from an uncalibrated camera. Our method improves generalization in 31 out of 36 environments across various tasks and outperforms domain randomization on a majority of environments. Webpage and implementation: https://nicklashansen.github.io/PAD/.

ICLR Conference 2021 Conference Paper

What Should Not Be Contrastive in Contrastive Learning

• Tete Xiao
• Xiaolong Wang 0004
• Alexei A. Efros
• Trevor Darrell

Recent self-supervised contrastive methods have been able to produce impressive transferable visual representations by learning to be invariant to different data augmentations. However, these methods implicitly assume a particular set of representational invariances (e.g., invariance to color), and can perform poorly when a downstream task violates this assumption (e.g., distinguishing red vs. yellow cars). We introduce a contrastive learning framework which does not require prior knowledge of specific, task-dependent invariances. Our model learns to capture varying and invariant factors for visual representations by constructing separate embedding spaces, each of which is invariant to all but one augmentation. We use a multi-head network with a shared backbone which captures information across each augmentation and alone outperforms all baselines on downstream tasks. We further find that the concatenation of the invariant and varying spaces performs best across all tasks we investigate, including coarse-grained, fine-grained, and few-shot downstream classification tasks, and various data corruptions.

ICML Conference 2020 Conference Paper

Test-Time Training with Self-Supervision for Generalization under Distribution Shifts

• Yu Sun 0020
• Xiaolong Wang 0004
• Zhuang Liu 0003
• John Miller 0001
• Alexei A. Efros
• Moritz Hardt

In this paper, we propose Test-Time Training, a general approach for improving the performance of predictive models when training and test data come from different distributions. We turn a single unlabeled test sample into a self-supervised learning problem, on which we update the model parameters before making a prediction. This also extends naturally to data in an online stream. Our simple approach leads to improvements on diverse image classification benchmarks aimed at evaluating robustness to distribution shifts.

ICML Conference 2018 Conference Paper

CyCADA: Cycle-Consistent Adversarial Domain Adaptation

• Judy Hoffman
• Eric Tzeng
• Taesung Park
• Jun-Yan Zhu
• Phillip Isola
• Kate Saenko
• Alexei A. Efros
• Trevor Darrell

Domain adaptation is critical for success in new, unseen environments. Adversarial adaptation models have shown tremendous progress towards adapting to new environments by focusing either on discovering domain invariant representations or by mapping between unpaired image domains. While feature space methods are difficult to interpret and sometimes fail to capture pixel-level and low-level domain shifts, image space methods sometimes fail to incorporate high level semantic knowledge relevant for the end task. We propose a model which adapts between domains using both generative image space alignment and latent representation space alignment. Our approach, Cycle-Consistent Adversarial Domain Adaptation (CyCADA), guides transfer between domains according to a specific discriminatively trained task and avoids divergence by enforcing consistency of the relevant semantics before and after adaptation. We evaluate our method on a variety of visual recognition and prediction settings, including digit classification and semantic segmentation of road scenes, advancing state-of-the-art performance for unsupervised adaptation from synthetic to real world driving domains.

ICML Conference 2018 Conference Paper

Investigating Human Priors for Playing Video Games

• Rachit Dubey
• Pulkit Agrawal 0001
• Deepak Pathak
• Thomas L. Griffiths 0001
• Alexei A. Efros

What makes humans so good at solving seemingly complex video games? Unlike computers, humans bring in a great deal of prior knowledge about the world, enabling efficient decision making. This paper investigates the role of human priors for solving video games. Given a sample game, we conduct a series of ablation studies to quantify the importance of various priors on human performance. We do this by modifying the video game environment to systematically mask different types of visual information that could be used by humans as priors. We find that removal of some prior knowledge causes a drastic degradation in the speed with which human players solve the game, e. g. from 2 minutes to over 20 minutes. Furthermore, our results indicate that general priors, such as the importance of objects and visual consistency, are critical for efficient game-play. Videos and the game manipulations are available at https: //rach0012. github. io/humanRL_website/

ICLR Conference 2018 Conference Paper

Zero-Shot Visual Imitation

• Deepak Pathak
• Parsa Mahmoudieh
• Guanghao Luo
• Pulkit Agrawal 0001
• Dian Chen 0001
• Yide Shentu
• Evan Shelhamer
• Jitendra Malik

The current dominant paradigm for imitation learning relies on strong supervision of expert actions to learn both 'what' and 'how' to imitate. We pursue an alternative paradigm wherein an agent first explores the world without any expert supervision and then distills its experience into a goal-conditioned skill policy with a novel forward consistency loss. In our framework, the role of the expert is only to communicate the goals (i.e., what to imitate) during inference. The learned policy is then employed to mimic the expert (i.e., how to imitate) after seeing just a sequence of images demonstrating the desired task. Our method is 'zero-shot' in the sense that the agent never has access to expert actions during training or for the task demonstration at inference. We evaluate our zero-shot imitator in two real-world settings: complex rope manipulation with a Baxter robot and navigation in previously unseen office environments with a TurtleBot. Through further experiments in VizDoom simulation, we provide evidence that better mechanisms for exploration lead to learning a more capable policy which in turn improves end task performance. Videos, models, and more details are available at https://pathak22.github.io/zeroshot-imitation/.

ICML Conference 2017 Conference Paper

Curiosity-driven Exploration by Self-supervised Prediction

• Deepak Pathak
• Pulkit Agrawal 0001
• Alexei A. Efros
• Trevor Darrell

In many real-world scenarios, rewards extrinsic to the agent are extremely sparse, or absent altogether. In such cases, curiosity can serve as an intrinsic reward signal to enable the agent to explore its environment and learn skills that might be useful later in its life. We formulate curiosity as the error in an agent’s ability to predict the consequence of its own actions in a visual feature space learned by a self-supervised inverse dynamics model. Our formulation scales to high-dimensional continuous state spaces like images, bypasses the difficulties of directly predicting pixels, and, critically, ignores the aspects of the environment that cannot affect the agent. The proposed approach is evaluated in two environments: VizDoom and Super Mario Bros. Three broad settings are investigated: 1) sparse extrinsic reward, where curiosity allows for far fewer interactions with the environment to reach the goal; 2) exploration with no extrinsic reward, where curiosity pushes the agent to explore more efficiently; and 3) generalization to unseen scenarios (e. g. new levels of the same game) where the knowledge gained from earlier experience helps the agent explore new places much faster than starting from scratch.

ICML Conference 2012 Conference Paper

Exemplar-SVMs for Visual Ob ject Detection, Label Transfer and Image Retrieval

• Tomasz Malisiewicz
• Abhinav Shrivastava
• Abhinav Gupta 0001
• Alexei A. Efros

IROS Conference 2006 Conference Paper

Opportunistic Use of Vision to Push Back the Path-Planning Horizon

• Bart C. Nabbe
• Derek Hoiem
• Alexei A. Efros
• Martial Hebert

Mobile robots need maps or other forms of geometric information about the environment to navigate. The mobility sensors (LADAR, stereo, etc.) on these robotic vehicles can however populate these maps only up to a distance of a few tens of meters. A navigation system has no knowledge about the world beyond this sensing horizon. As a result, path planners that rely only on this knowledge are unable to anticipate obstacles sufficiently early and have no choice but to resort to an inefficient local obstacle avoidance behavior. However, recent developments in the computer vision community allows us to collect geometric information about the environment far beyond this sensing horizon. The coarse 3D geometric estimation that can be recovered is derived from an appearance-based model. That uses a multiple-hypothesis framework to robustly estimate scene structure from a single image and estimating confidences for each geometric label. This 3D geometric estimation is used with a previously presented navigation strategy that reasons about sensor constraints and plans for measurements while navigating towards the goal. The validity of the sensing method and navigation strategy is supported by results from simulations as well as field experiments with a real robotic platform. These results also show that significant reduction in path length can be achieved by using this framework