Generating Chain-of-Thought (CoT) before deriving the answer can effectively improve the reasoning capabilities of large language models (LLMs) and significantly improve the accuracy of the generated answer. However, in most cases, the length of the generated CoT is much longer than the desired final answer, which results in additional decoding costs. Furthermore, existing research has discovered that shortening the reasoning steps in CoT, even while preserving the key information, diminishes LLMs' abilities. These phenomena make it difficult to use LLMs and CoT in many real-world applications that only require the final answer and are sensitive to latency, such as search and recommendation. To reduce the costs of model decoding and shorten the length of the generated CoT, this paper presents Conditioned Compressed Chain-of-Thought (C3oT), a CoT compression framework that involves a compressor to compress an original longer CoT into a shorter CoT while maintaining key information and interpretability, a conditioned training method to train LLMs with both longer CoT and shorter CoT simultaneously to learn the corresponding relationships between them, and a conditioned inference method to gain the reasoning ability learned from longer CoT by generating shorter CoT. We conduct experiments over four datasets from arithmetic and commonsense scenarios, showing that the proposed method is capable of compressing the length of generated CoT by up to more than 50% without compromising its effectiveness.

In recent years, many researchers have leveraged structural information from dependency trees to improve Named Entity Recognition (NER). Most of their methods take dependency-tree labels as input features for NER model training. However, such dependency information is not inherently provided in most NER corpora, making the methods with low usability in practice. To effectively exploit the potential of word-dependency knowledge, motivated by the success of Multi-Task Learning on cross-domain NER, we investigate a novel NER learning method incorporating cross-domain Dependency Parsing (DP) as its auxiliary learning task. Then, considering the high consistency of word-dependency relations across domains, we present an unsupervised domain-adapted method to transfer word-dependency knowledge from high-resource domains to low-resource ones. With the help of cross-domain DP to bridge different domains, both useful cross-domain and cross-task knowledge can be learned by our model to considerably benefit cross-domain NER. To make better use of the cross-task knowledge between NER and DP, we unify both tasks in a shared network architecture for joint learning, using Maximum Mean Discrepancy(MMD). Finally, through extensive experiments, we show our proposed method can not only effectively take advantage of word-dependency knowledge, but also significantly outperform other Multi-Task Learning methods on cross-domain NER. Our code is open-source and available at https://github.com/xianghuisun/DADP.

Natural image statistics have been widely exploited for lossless predictive coding and other applications. However, traditional adaptive techniques always focus on the local consistency of training set regardless of what the predicted target looks like. We investigate the problem of introducing the model evidence of predicted target since self-similarity inherent in natural images gives some kind of prior information for the distribution of predicted result. The proposed Bayesian model integrated with both training evidence and target evidence takes full advantages of local structure as well as self-similarity. Experimental results demonstrate that the proposed context model achieves best results compared with the state-of-the-art lossless predictors.

For current P2P-VoD systems, three fundamental problems exit in user experience: exceedingly large startup delay, long jump latency, and poor playback continuity. These problems primarily stem from lack of media data. In this paper, we propose Multi-Task Downloading with Bandwidth Control (MTD(BC)), an efficient and practical mechanism to prefetch media data. In MTD, a user can download multiple videos in parallel with its current viewing, which significantly decreases video switching delays. However, MTD brings a serious problem: downloading "other" tasks could impede the playback performance of the current viewing, especially in low-bandwidth network. This problem is solved through our design of bandwidth control. To our knowledge, we are the first to propose MTD with bandwidth control for P2P-VoD and conduct empirical evaluations in the real-world system. The running results show that MTD(BC) achieves better streaming quality than the traditional method. In particular, our mechanism reduces 75% of startup delay and 36% of jump latency in low-bandwidth network with high system scalability.