Ming Wu

Xuandong Zhao, Siqi Ouyang, Zhiguo Yu et al. · berkeley, cmu

How can we extend a pre-trained model to many language understanding tasks, without labeled or additional unlabeled data? Pre-trained language models (PLMs) have been effective for a wide range of NLP tasks. However, existing approaches either require fine-tuning on downstream labeled datasets or manually constructing proper prompts. In this paper, we propose nonparametric prompting PLM (NPPrompt) for fully zero-shot language understanding. Unlike previous methods, NPPrompt uses only pre-trained language models and does not require any labeled data or additional raw corpus for further fine-tuning, nor does it rely on humans to construct a comprehensive set of prompt label words. We evaluate NPPrompt against previous major few-shot and zero-shot learning methods on diverse NLP tasks: including text classification, text entailment, similar text retrieval, and paraphrasing. Experimental results demonstrate that our NPPrompt outperforms the previous best fully zero-shot method by big margins, with absolute gains of 12.8% in accuracy on text classification and 18.9% on the GLUE benchmark.

Xuandong Zhao, Zhiguo Yu, Ming Wu et al. · berkeley, cmu

How to learn highly compact yet effective sentence representation? Pre-trained language models have been effective in many NLP tasks. However, these models are often huge and produce large sentence embeddings. Moreover, there is a big performance gap between large and small models. In this paper, we propose Homomorphic Projective Distillation (HPD) to learn compressed sentence embeddings. Our method augments a small Transformer encoder model with learnable projection layers to produce compact representations while mimicking a large pre-trained language model to retain the sentence representation quality. We evaluate our method with different model sizes on both semantic textual similarity (STS) and semantic retrieval (SR) tasks. Experiments show that our method achieves 2.7-4.5 points performance gain on STS tasks compared with previous best representations of the same size. In SR tasks, our method improves retrieval speed (8.2$\times$) and memory usage (8.0$\times$) compared with state-of-the-art large models.

Zhenduo Wang, Yuancheng Tu, Corby Rosset et al. · tsinghua

A long-standing challenge for search and conversational assistants is query intention detection in ambiguous queries. Asking clarifying questions in conversational search has been widely studied and considered an effective solution to resolve query ambiguity. Existing work have explored various approaches for clarifying question ranking and generation. However, due to the lack of real conversational search data, they have to use artificial datasets for training, which limits their generalizability to real-world search scenarios. As a result, the industry has shown reluctance to implement them in reality, further suspending the availability of real conversational search interaction data. The above dilemma can be formulated as a cold start problem of clarifying question generation and conversational search in general. Furthermore, even if we do have large-scale conversational logs, it is not realistic to gather training data that can comprehensively cover all possible queries and topics in open-domain search scenarios. The risk of fitting bias when training a clarifying question retrieval/generation model on incomprehensive dataset is thus another important challenge. In this work, we innovatively explore generating clarifying questions in a zero-shot setting to overcome the cold start problem and we propose a constrained clarifying question generation system which uses both question templates and query facets to guide the effective and precise question generation. The experiment results show that our method outperforms existing state-of-the-art zero-shot baselines by a large margin. Human annotations to our model outputs also indicate our method generates 25.2\% more natural questions, 18.1\% more useful questions, 6.1\% less unnatural and 4\% less useless questions.

Shizun Wang, Weihong Zeng, Xu Wang et al.

The creation of a parameterized stylized character involves careful selection of numerous parameters, also known as the "avatar vectors" that can be interpreted by the avatar engine. Existing unsupervised avatar vector estimation methods that auto-create avatars for users, however, often fail to work because of the domain gap between realistic faces and stylized avatar images. To this end, we propose SwiftAvatar, a novel avatar auto-creation framework that is evidently superior to previous works. SwiftAvatar introduces dual-domain generators to create pairs of realistic faces and avatar images using shared latent codes. The latent codes can then be bridged with the avatar vectors as pairs, by performing GAN inversion on the avatar images rendered from the engine using avatar vectors. Through this way, we are able to synthesize paired data in high-quality as many as possible, consisting of avatar vectors and their corresponding realistic faces. We also propose semantic augmentation to improve the diversity of synthesis. Finally, a light-weight avatar vector estimator is trained on the synthetic pairs to implement efficient auto-creation. Our experiments demonstrate the effectiveness and efficiency of SwiftAvatar on two different avatar engines. The superiority and advantageous flexibility of SwiftAvatar are also verified in both subjective and objective evaluations.

Banghua Zhu, Mingyu Ding, Philip Jacobson et al.

Self-training is an important technique for solving semi-supervised learning problems. It leverages unlabeled data by generating pseudo-labels and combining them with a limited labeled dataset for training. The effectiveness of self-training heavily relies on the accuracy of these pseudo-labels. In this paper, we introduce doubly robust self-training, a novel semi-supervised algorithm that provably balances between two extremes. When the pseudo-labels are entirely incorrect, our method reduces to a training process solely using labeled data. Conversely, when the pseudo-labels are completely accurate, our method transforms into a training process utilizing all pseudo-labeled data and labeled data, thus increasing the effective sample size. Through empirical evaluations on both the ImageNet dataset for image classification and the nuScenes autonomous driving dataset for 3D object detection, we demonstrate the superiority of the doubly robust loss over the standard self-training baseline.

Cheng Lyu, Jiake Xie, Bo Xu et al.

Performance of trimap-free image matting methods is limited when trying to decouple the deterministic and undetermined regions, especially in the scenes where foregrounds are semantically ambiguous, chromaless, or high transmittance. In this paper, we propose a novel framework named Privileged Prior Information Distillation for Image Matting (PPID-IM) that can effectively transfer privileged prior environment-aware information to improve the performance of students in solving hard foregrounds. The prior information of trimap regulates only the teacher model during the training stage, while not being fed into the student network during actual inference. In order to achieve effective privileged cross-modality (i.e. trimap and RGB) information distillation, we introduce a Cross-Level Semantic Distillation (CLSD) module that reinforces the trimap-free students with more knowledgeable semantic representations and environment-aware information. We also propose an Attention-Guided Local Distillation module that efficiently transfers privileged local attributes from the trimap-based teacher to trimap-free students for the guidance of local-region optimization. Extensive experiments demonstrate the effectiveness and superiority of our PPID framework on the task of image matting. In addition, our trimap-free IndexNet-PPID surpasses the other competing state-of-the-art methods by a large margin, especially in scenarios with chromaless, weak texture, or irregular objects.

Haotian Yan, Chuang Zhang, Ming Wu

Multi-scale representations are crucial for semantic segmentation. The community has witnessed the flourish of semantic segmentation convolutional neural networks (CNN) exploiting multi-scale contextual information. Motivated by that the vision transformer (ViT) is powerful in image classification, some semantic segmentation ViTs are recently proposed, most of them attaining impressive results but at a cost of computational economy. In this paper, we succeed in introducing multi-scale representations into semantic segmentation ViT via window attention mechanism and further improves the performance and efficiency. To this end, we introduce large window attention which allows the local window to query a larger area of context window at only a little computation overhead. By regulating the ratio of the context area to the query area, we enable the $\textit{large window attention}$ to capture the contextual information at multiple scales. Moreover, the framework of spatial pyramid pooling is adopted to collaborate with $\textit{the large window attention}$, which presents a novel decoder named $\textbf{la}$rge $\textbf{win}$dow attention spatial pyramid pooling (LawinASPP) for semantic segmentation ViT. Our resulting ViT, Lawin Transformer, is composed of an efficient hierachical vision transformer (HVT) as encoder and a LawinASPP as decoder. The empirical results demonstrate that Lawin Transformer offers an improved efficiency compared to the existing method. Lawin Transformer further sets new state-of-the-art performance on Cityscapes (84.4% mIoU), ADE20K (56.2% mIoU) and COCO-Stuff datasets. The code will be released at https://github.com/yan-hao-tian/lawin

Ji Qi, WenPeng Zhu, Li Li et al.

The distributed training of foundation models, particularly large language models (LLMs), demands a high level of communication. Consequently, it is highly dependent on a centralized cluster with fast and reliable interconnects. Can we conduct training on slow networks and thereby unleash the power of decentralized clusters when dealing with models exceeding 100 billion parameters? In this paper, we propose DiLoCoX, a low-communication large-scale decentralized cluster training framework. It combines Pipeline Parallelism with Dual Optimizer Policy, One-Step-Delay Overlap of Communication and Local Training, and an Adaptive Gradient Compression Scheme. This combination significantly improves the scale of parameters and the speed of model pre-training. We justify the benefits of one-step-delay overlap of communication and local training, as well as the adaptive gradient compression scheme, through a theoretical analysis of convergence. Empirically, we demonstrate that DiLoCoX is capable of pre-training a 107B foundation model over a 1Gbps network. Compared to vanilla AllReduce, DiLoCoX can achieve a 357x speedup in distributed training while maintaining negligible degradation in model convergence. To the best of our knowledge, this is the first decentralized training framework successfully applied to models with over 100 billion parameters.