We consider the issue of calibration in large language models (LLM). Recent studies have found that common interventions such as instruction tuning often result in poorly calibrated LLMs. Although calibration is well-...
详细信息
We consider the issue of calibration in large language models (LLM). Recent studies have found that common interventions such as instruction tuning often result in poorly calibrated LLMs. Although calibration is well-explored in traditional applications, calibrating LLMs is uniquely challenging. These challenges stem as much from the severe computational requirements of LLMs as from their versatility, which allows them to be applied to diverse tasks. Addressing these challenges, we propose THERMOMETER, a calibration approach tailored to LLMs. THERMOMETER learns an auxiliary model, given data from multiple tasks, for calibrating a LLM. It is computationally efficient, preserves the accuracy of the LLM, and produces better-calibrated responses for new tasks. Extensive empirical evaluations across various benchmarks demonstrate the effectiveness of the proposed method. Copyright 2024 by the author(s)
Videos are becoming a ubiquitous means of sharing information on social media platforms. In response, data videos—short clips combining visualization with dynamic storytelling, audio descriptions, and spatial referen...
详细信息
Device-to-device (D2D) communication, in contrast to typical cellular connections, is a direct connection between adjacent mobile users;it skips the base station (BS) and does not rely on network infrastructure to del...
详细信息
TOTEM is a framework that allows users to execute their own code on access restricted datasets with controlled computation. It provides data security by restricting the exchange of data across the network, instead, th...
详细信息
One significant advantage of microservice cloud architectures is the agility with which microservices may be duplicated to enhance overall service quality and satisfy service-level contracts. However, the complex inte...
详细信息
Recently, it has been demonstrated that a solution set that is better than the final population can be obtained by subset selection in some studies on evolutionary multi-objective optimization. The main challenge in t...
详细信息
Security systems are the need of the hour to protect data from unauthorized *** dissemination of confidential information over the public network requires a high level of *** security approach such as steganography en...
详细信息
Security systems are the need of the hour to protect data from unauthorized *** dissemination of confidential information over the public network requires a high level of *** security approach such as steganography ensures confidentiality,authentication,integrity,and *** helps in hiding the secret data inside the cover media so that the attacker can be confused during the transmission process of secret data between sender and ***,we present an efficient hybrid security model that provides multifold security *** this end,a rectified Advanced Encryption Standard(AES)algorithm is proposed to overcome the problems existing in AES such as pattern appearance and high *** modified AES is used for the encryption of the stego image that contains the digitally signed encrypted secret *** enciphering and deciphering of the secret data are done using the Rivest–Shamir–Adleman(RSA)*** experiments are conducted on the images of the USC-SIPI standard image *** experimental results prove that the proposed hybrid system outperforms other SOTA(state-of-the-art)approaches.
The rapid development of Low Earth Orbit (LEO) satellite networks has provided ubiquitous Internet access to users around the world, especially in areas where there are no terrestrial networks. However, a dish can onl...
详细信息
A common issue in learning decision-making policies in data-rich settings is spurious correlations in the offline dataset, which can be caused by hidden confounders. Instrumental variable (IV) regression, which utilis...
A common issue in learning decision-making policies in data-rich settings is spurious correlations in the offline dataset, which can be caused by hidden confounders. Instrumental variable (IV) regression, which utilises a key unconfounded variable known as the instrument, is a standard technique for learning causal relationships between confounded action, outcome, and context variables. Most recent IV regression algorithms use a two-stage approach, where a deep neural network (DNN) estimator learnt in the first stage is directly plugged into the second stage, in which another DNN is used to estimate the causal effect. Naively plugging the estimator can cause heavy bias in the second stage, especially when regularisation bias is present in the first stage estimator. We propose DML-IV, a non-linear IV regression method that reduces the bias in two-stage IV regressions and effectively learns high-performing policies. We derive a novel learning objective to reduce bias and design the DML-IV algorithm following the double/debiased machine learning (DML) framework. The learnt DML-IV estimator has strong convergence rate and O(N-1/2) suboptimality guarantees that match those when the dataset is unconfounded. DML-IV outperforms state-of-the-art IV regression methods on IV regression benchmarks and learns high-performing policies in the presence of instruments.
The rapid advancement of Internet of Things (IoT) technology has transformed healthcare by enabling remote patient monitoring, chronic disease management, and preventive care solutions. Human Body Communication (HBC) ...
详细信息
暂无评论