Learning from human feedback has been shown to improve text-to-image models. These techniques first learn a reward function that captures what humans care about in the task and then improve the models based on the learned reward function. Even though relatively simple approaches (e.g., rejection sampling based on reward scores) have been investigated, fine-tuning text-to-image models with the reward function remains challenging. In this work, we propose using online reinforcement learning (RL) to fine-tune text-to-image models. We focus on diffusion models, defining the fine-tuning task as an RL problem, and updating the pre-trained text-to-image diffusion models using policy gradient to maximize the feedback-trained reward. Our approach, coined DPOK, integrates policy optimization with KL regularization. We conduct an analysis of KL regularization for both RL fine-tuning and supervised fine-tuning. In our experiments, we show that DPOK is generally superior to supervised fine-tuning with respect to both image-text alignment and image quality. Our code is available at //github.com/google-research/google-research/tree/master/dpok.
相關內容
ACM/IEEE第23屆模型驅動工程語言和系統國際會議，是模型驅動軟件和系統工程的首要會議系列，由ACM-SIGSOFT和IEEE-TCSE支持組織。自1998年以來，模型涵蓋了建模的各個方面，從語言和方法到工具和應用程序。模特的參加者來自不同的背景，包括研究人員、學者、工程師和工業專業人士。MODELS 2019是一個論壇，參與者可以圍繞建模和模型驅動的軟件和系統交流前沿研究成果和創新實踐經驗。今年的版本將為建模社區提供進一步推進建模基礎的機會，并在網絡物理系統、嵌入式系統、社會技術系統、云計算、大數據、機器學習、安全、開源等新興領域提出建模的創新應用以及可持續性。
官網鏈接： ·
泛函
·
有向
·
FAST
·
3D
·
Neural 3D implicit representations learn priors that are useful for diverse applications, such as single- or multiple-view 3D reconstruction. A major downside of existing approaches while rendering an image is that they require evaluating the network multiple times per camera ray so that the high computational time forms a bottleneck for downstream applications. We address this problem by introducing a novel neural scene representation that we call the directional distance function (DDF). To this end, we learn a signed distance function (SDF) along with our DDF model to represent a class of shapes. Specifically, our DDF is defined on the unit sphere and predicts the distance to the surface along any given direction. Therefore, our DDF allows rendering images with just a single network evaluation per camera ray. Based on our DDF, we present a novel fast algorithm (FIRe) to reconstruct 3D shapes given a posed depth map. We evaluate our proposed method on 3D reconstruction from single-view depth images, where we empirically show that our algorithm reconstructs 3D shapes more accurately and it is more than 15 times faster (per iteration) than competing methods.
We introduce novel methods for state estimation, feedforward and feedback control, which specifically target humanoid robots with hardware limitations. Our method combines a five-mass model with approximate dynamics of each mass. It enables acquiring an accurate assessment of the centroidal state and Center of Pressure, even when direct forms of force or contact sensing are unavailable. Upon this, we develop a feedforward scheme that operates on the centroidal state, accounting for insufficient joint tracking capabilities. Finally, we implement feedback mechanisms, which compensate for the lack in Degrees of Freedom that our NimbRo-OP2X robot has. The whole approach allows for reactive stepping to maintain balance despite these limitations, which was verified on hardware during RoboCup 2023, in Bordeaux, France.
Despite the impressive results of arbitrary image-guided style transfer methods, text-driven image stylization has recently been proposed for transferring a natural image into a stylized one according to textual descriptions of the target style provided by the user. Unlike the previous image-to-image transfer approaches, text-guided stylization progress provides users with a more precise and intuitive way to express the desired style. However, the huge discrepancy between cross-modal inputs/outputs makes it challenging to conduct text-driven image stylization in a typical feed-forward CNN pipeline. In this paper, we present DiffStyler, a dual diffusion processing architecture to control the balance between the content and style of the diffused results. The cross-modal style information can be easily integrated as guidance during the diffusion process step-by-step. Furthermore, we propose a content image-based learnable noise on which the reverse denoising process is based, enabling the stylization results to better preserve the structure information of the content image. We validate the proposed DiffStyler beyond the baseline methods through extensive qualitative and quantitative experiments. Code is available at \url{//github.com/haha-lisa/Diffstyler}.
Open-loop stable limit cycles are foundational to the dynamics of legged robots. They impart a self-stabilizing character to the robot's gait, thus alleviating the need for compute-heavy feedback-based gait correction. This paper proposes a general approach to rapidly generate limit cycles with explicit stability constraints for a given dynamical system. In particular, we pose the problem of open-loop limit cycle stability as a single-stage constrained-optimization problem (COP), and use Direct Collocation to transcribe it into a nonlinear program (NLP) with closed-form expressions for constraints, objectives, and their gradients. The COP formulations of stability are developed based (1) on the spectral radius of a discrete return map, and (2) on the spectral radius of the system's monodromy matrix, where the spectral radius is bounded using different constraint-satisfaction formulations of the eigenvalue problem. We compare the performance and solution qualities of each approach, but specifically highlight the Schur decomposition of the monodromy matrix as a formulation which boasts wider applicability through weaker assumptions and attractive numerical convergence properties. Moreover, we present results from our experiments on a spring-loaded inverted pendulum model of a robot, where our method generated actuation trajectories for open-loop stable hopping in under 2 seconds (on the Intel Core i7-6700K), and produced energy-minimizing actuation trajectories even under tight stability constraints.
Articulated object manipulation is a fundamental yet challenging task in robotics. Due to significant geometric and semantic variations across object categories, previous manipulation models struggle to generalize to novel categories. Few-shot learning is a promising solution for alleviating this issue by allowing robots to perform a few interactions with unseen objects. However, extant approaches often necessitate costly and inefficient test-time interactions with each unseen instance. Recognizing this limitation, we observe that despite their distinct shapes, different categories often share similar local geometries essential for manipulation, such as pullable handles and graspable edges - a factor typically underutilized in previous few-shot learning works. To harness this commonality, we introduce 'Where2Explore', an affordance learning framework that effectively explores novel categories with minimal interactions on a limited number of instances. Our framework explicitly estimates the geometric similarity across different categories, identifying local areas that differ from shapes in the training categories for efficient exploration while concurrently transferring affordance knowledge to similar parts of the objects. Extensive experiments in simulated and real-world environments demonstrate our framework's capacity for efficient few-shot exploration and generalization.
Process mining, a data-driven approach for analyzing, visualizing, and improving business processes using event logs, has emerged as a powerful technique in the field of business process management. Process forecasting is a sub-field of process mining that studies how to predict future processes and process models. In this paper, we introduce and motivate the problem of event log prediction and present our approach to solving the event log prediction problem, in particular, using the sequence-to-sequence deep learning approach. We evaluate and analyze the prediction outcomes on a variety of synthetic logs and seven real-life logs and show that our approach can generate perfect predictions on synthetic logs and that deep learning techniques have the potential to be applied in real-world event log prediction tasks. We further provide practical recommendations for event log predictions grounded in the outcomes of the conducted experiments.
Widely used alignment techniques, such as reinforcement learning from human feedback (RLHF), rely on the ability of humans to supervise model behavior - for example, to evaluate whether a model faithfully followed instructions or generated safe outputs. However, future superhuman models will behave in complex ways too difficult for humans to reliably evaluate; humans will only be able to weakly supervise superhuman models. We study an analogy to this problem: can weak model supervision elicit the full capabilities of a much stronger model? We test this using a range of pretrained language models in the GPT-4 family on natural language processing (NLP), chess, and reward modeling tasks. We find that when we naively finetune strong pretrained models on labels generated by a weak model, they consistently perform better than their weak supervisors, a phenomenon we call weak-to-strong generalization. However, we are still far from recovering the full capabilities of strong models with naive finetuning alone, suggesting that techniques like RLHF may scale poorly to superhuman models without further work. We find that simple methods can often significantly improve weak-to-strong generalization: for example, when finetuning GPT-4 with a GPT-2-level supervisor and an auxiliary confidence loss, we can recover close to GPT-3.5-level performance on NLP tasks. Our results suggest that it is feasible to make empirical progress today on a fundamental challenge of aligning superhuman models.
Diffusion models (DMs) have shown great potential for high-quality image synthesis. However, when it comes to producing images with complex scenes, how to properly describe both image global structures and object details remains a challenging task. In this paper, we present Frido, a Feature Pyramid Diffusion model performing a multi-scale coarse-to-fine denoising process for image synthesis. Our model decomposes an input image into scale-dependent vector quantized features, followed by a coarse-to-fine gating for producing image output. During the above multi-scale representation learning stage, additional input conditions like text, scene graph, or image layout can be further exploited. Thus, Frido can be also applied for conditional or cross-modality image synthesis. We conduct extensive experiments over various unconditioned and conditional image generation tasks, ranging from text-to-image synthesis, layout-to-image, scene-graph-to-image, to label-to-image. More specifically, we achieved state-of-the-art FID scores on five benchmarks, namely layout-to-image on COCO and OpenImages, scene-graph-to-image on COCO and Visual Genome, and label-to-image on COCO. Code is available at //github.com/davidhalladay/Frido.
The incredible development of federated learning (FL) has benefited various tasks in the domains of computer vision and natural language processing, and the existing frameworks such as TFF and FATE has made the deployment easy in real-world applications. However, federated graph learning (FGL), even though graph data are prevalent, has not been well supported due to its unique characteristics and requirements. The lack of FGL-related framework increases the efforts for accomplishing reproducible research and deploying in real-world applications. Motivated by such strong demand, in this paper, we first discuss the challenges in creating an easy-to-use FGL package and accordingly present our implemented package FederatedScope-GNN (FS-G), which provides (1) a unified view for modularizing and expressing FGL algorithms; (2) comprehensive DataZoo and ModelZoo for out-of-the-box FGL capability; (3) an efficient model auto-tuning component; and (4) off-the-shelf privacy attack and defense abilities. We validate the effectiveness of FS-G by conducting extensive experiments, which simultaneously gains many valuable insights about FGL for the community. Moreover, we employ FS-G to serve the FGL application in real-world E-commerce scenarios, where the attained improvements indicate great potential business benefits. We publicly release FS-G, as submodules of FederatedScope, at //github.com/alibaba/FederatedScope to promote FGL's research and enable broad applications that would otherwise be infeasible due to the lack of a dedicated package.
Distant supervision can effectively label data for relation extraction, but suffers from the noise labeling problem. Recent works mainly perform soft bag-level noise reduction strategies to find the relatively better samples in a sentence bag, which is suboptimal compared with making a hard decision of false positive samples in sentence level. In this paper, we introduce an adversarial learning framework, which we named DSGAN, to learn a sentence-level true-positive generator. Inspired by Generative Adversarial Networks, we regard the positive samples generated by the generator as the negative samples to train the discriminator. The optimal generator is obtained until the discrimination ability of the discriminator has the greatest decline. We adopt the generator to filter distant supervision training dataset and redistribute the false positive instances into the negative set, in which way to provide a cleaned dataset for relation classification. The experimental results show that the proposed strategy significantly improves the performance of distant supervision relation extraction comparing to state-of-the-art systems.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191