This work presents a novel and parameter-efficient transfer learning framework. The framework consists of two phases: knowledge modeling based on prompt decomposition and knowledge transfer based on attention. Specifically, during the first phase, we decompose the prompt into parameter spaces of different ranks and leverage their characteristics to precisely model both general knowledge and task-specific knowledge separately. During the second phase, we train an attention module to adaptively integrate task-specific knowledge and generate an instance-wise prompt, which is then further fine-tuned. Through these two stages, our approach can accurately and efficiently transfer task knowledge of different granularities and types based on the input sample during inference. Extensive experiments demonstrate that our approach outperforms the multi-task learning baselines and SOTA parameter-efficient transfer learning methods. Furthermore, despite using only 0.13% of the parameters compared to full-parameter fine-tuning, it achieves an absolute improvement ranging from 0.3 to 3.9 across different benchmarks.

Multi-unit auctions are a paradigmatic model of resource allocation , where a seller brings multiple units of a good to a set of buyers equipped with monetary budgets. It is well known that Walrasian equilibria do not always exist in this model, however compelling relaxations such as Walrasian envy-free pricing do. We design a best possible envy-free and prior-free mechanism for multi-unit auctions with budgets. When the market is even mildly competitive, the approximation ratios of this mechanism are small constants for both the revenue and welfare objectives, and in fact for welfare the approximation converges to 1 as the market becomes fully competitive. We also give an impossibility theorem , showing that truthfulness requires discarding resources and is thus incompatible with (Pareto) efficiency. • We consider multi-unit auctions with budgets and design a best possible envy-free and prior-free mechanism. • The mechanism approximates the optimal revenue and welfare within constant factors. • For welfare, the approximation converges to 1 as the market becomes fully competitive.

Blockchain is an important supporting technology for various sustainable systems. It relies on a number of distributed nodes running blockchain client software, which is responsible for some critical tasks, such as communicating with other nodes and generating new blocks. However, the quick evolution of blockchain technology brings crucial challenges to blockchain client design. After carefully examining existing blockchain client software, we have identified a critical weakness: Blockchain clients are weak in supporting live upgrades, resulting in a blockchain fork that incurs security concerns and risks. In this article, we propose Phoenix, a novel blockchain client design that is live upgradable. Phoenix uses blockchain service encapsulation to decouple blockchain services. Based on service encapsulation, we propose a live upgrade scheme that packs upgrade codes into blockchain transactions and uses a Just-In-Time engine to avoid service interruption. A parallel execution engine is developed to increase service efficiency. We evaluated Phoenix on a 51-node blockchain, and experimental results show that Phoenix outperforms existing solutions in overhead and upgrade latency.

In order to diagnostically analyze and improve the capability of pretrained language models (PLMs) in text generation, we propose TGEA 2.0, to date the largest dataset built on machine-authored texts by PLMs with fine-grained semantic annotations on a wide variety of pathological generation errors. We collect 170K nominal, phrasal and sentential prompts from 6M natural sentences in 3 domains. These prompts are fed into 4 generative PLMs with their best decoding strategy to generate paragraphs. 195,629 sentences are extracted from these generated paragraphs for manual annotation, where 36K erroneous sentences are detected, 42K erroneous spans are located and categorized into an error type defined in a two-level error taxonomy. We define a \textbf{Mi}nimal \textbf{S}et of \textbf{E}rror-related \textbf{W}ords (MiSEW) for each erroneous span, which not only provides error-associated words but also rationalizes the reasoning behind the error. Quality control with a pre-annotation and feedback loop is performed before and during the entire annotation process. With the diagnostically annotated dataset, we propose 5 diagnosis benchmark tasks (i.e., erroneous text detection, MiSEW extraction, erroneous span location and correction together with error type classification) and 2 pathology mitigation benchmark tasks (pairwise comparison and word prediction). Experiment results on these benchmark tasks demonstrate that TGEA 2.0 is a challenging dataset that could facilitate further research on automatic diagnosis and pathology mitigation over machine texts. The dataset will be publicly available at https://github.com/tjunlp-lab/TGEA/.

Cryptocurrency mining, which has attracted substantial attention recently is a game where miners cost mining power and compete for an amount of digital token. In this paper, we revisit this game under new settings that miners' maximal mining powers, i.e. capacities, are disparate. We show it has a unique pure Nash-equilibrium and derive the closed-form. In the PNE, miners fall into two types: half-hearted and all-out, the former of which owns more capacity than the latter. An all-out miner mines with her full capacity while a half-hearted miner does not. Compared with the desired equilibrium that every miner does her best, our result reveals that the disparity of capacity leads to less total mining power, compromising the security of cryptocurrency systems.

We study the computation contests where players compete for searching a solution to a given problem with a winner-take-all reward. The search processes are independent across the players and the search speeds of players are proportional to their computational powers. One concrete application of this abstract model is the mining process of proof-of-work type blockchain systems, such as Bitcoin. Although one's winning probability is believed to be proportional to his computational power in previous studies on Bitcoin, we show that it is not the case in the strict sense.

In the standard form of mechanism design, a key assumption is that the designer has reliable information and technology to determine a prior distribution over types of the agents. In the meanwhile, as pointed out by the Wilson's Principle, a mechanism should rely as little as possible on the prior type distribution. In this paper, we put forward a simple model to formalize this statement. In our model, each agent has a true type distribution, according to which his type is drawn. In addition, the agent is able to commit to a fake type distribution and bids rationally as if his type were from the fake distribution (e.g., plays a Bayes equilibrium under the fake distributions). We investigate the equilibria of the induced distribution-reporting games among bidders, under the context of single-item auctions. We obtain several interesting findings: (1) the game induced by Myerson auction under our model is strategically equivalent to the first-price auction under the standard model. Consequently, the two games are revenue-equivalent. (2) the second-price auction, a well known prior independent auction, yields (weakly) more revenue than several reserve-based and virtual-value-based truthful, prior-dependent auctions, under our model. Our results complement the current literature which aims to show the superiority of prior-independent mechanisms.

In this paper, we analyze an emerging economic form, called fans economy, in which a fan donates money to the host and gets allocated proportional to the amount of his donation (normalized by the overall amount of donation). Fans economy is the major way live streaming apps monetize and includes a number of popular economic forms ranging from crowdfunding to mutual fund. We propose an auction game, coined all-pay auctions with proportional allocation (APAPA), to model the fans economy and analyze the auction from the perspective of revenue. Comparing to the standard all-pay auction, which normally has no pure Nash-Equilibrium in the complete information setting, we solve the pure Nash-Equilibrium of the APAPA in closed form and prove its uniqueness. Motivated by practical concerns, we then analyze the case where APAPA is equipped with a reserve and show that there might be multiple equilibria in this case. We give an efficient algorithm to compute all equilibria in this case. For either case, with or without reserve, we show that APAPA always extracts revenue that 2-approximates the second-highest valuation. Furthermore, we conduct experiments to show how revenue changes with respect to different reserves.

Multi-unit auctions are a paradigmatic model, where a seller brings multiple units of a good, while several buyers bring monetary endowments. It is well known that Walrasian equilibria do not always exist in this model, however compelling relaxations such as Walrasian envy-free pricing do. In this paper we design an optimal envy-free mechanism for multi-unit auctions with budgets. When the market is even mildly competitive, the approximation ratios of this mechanism are small constants for both the revenue and welfare objectives, and in fact for welfare the approximation converges to 1 as the market becomes fully competitive. We also give an impossibility theorem, showing that truthfulness requires discarding resources, and in particular, is incompatible with (Pareto) efficiency.

To realize the potential benefits of SDN, tremendous amount of network information has to be collected, stored, processed and retrieved for the central routing and resource allocation optimization in near real-time. However, the current SDN-Open Flow controller and management applications did not address this issue properly and could not easily and efficiently retrieve and utilize the available analytics on-time. Their "network information bases" are rudimentary, and could not facilitate the required fast processing and accessing for required intelligent information for large networks. We present an innovative mechanism that architects such information into a multi-dimensional model, utilizing prevailing technologies in large scale real-time data processing, the advances in in-memory SQL and NoSQL and the Online Analytical Processing (OLAP). Data and their hidden characteristics and inter-relationships can then be stored in-memory or cached using key/values as in NoSQL, while more static information could be persisted into local file system or SQL database, and globally in a centralized SQL database (or a data warehouse) that can store large amounts of historical data that can be easily processed with OLAP or data mining for intelligence. Extensive experiments and comparison with traditional relational database or in-memory hashing table clearly demonstrated the feasibility, capability and performance advantage of the proposed approach.