A large number of real and abstract systems involve the transformation of some basic resource into respective products under the action of multiple processing agents, which can be understood as multiple-agent production systems (MAP). At each discrete time instant, for each agent, a fraction of the resources is assumed to be kept, forwarded to other agents, or converted into work with some efficiency. The present work describes a systematic study of nine basic MAP architectures subdivided into two main groups, namely parallel and sequential distribution of resources from a single respective source. Several types of interconnections among the involved processing agents are also considered. The resulting MAP architectures are studied in terms of the total amount of work, the dispersion of the resources (states) among the agents, and the transition times from the start of operation until the respective steady state. Several interesting results are obtained and discussed, including the observation that some of the parallel designs were able to yield maximum work and minimum state dispersion, achieved at the expense of the transition time and use of several interconnections between the source and the agents. The results obtained for the sequential designs indicate that relatively high performance can be obtained for some specific cases.

Diagnosis of breast cancer malignancy at the early stages is a crucial step for controlling its side effects. Histopathological analysis provides a unique opportunity for malignant breast cancer detection. However, such a task would be tedious and time-consuming for the histopathologists. Deep Neural Networks enable us to learn informative features directly from raw histopathological images without manual feature extraction. Although Convolutional Neural Networks (CNNs) have been the dominant architectures in the computer vision realm, Transformer-based architectures have shown promising results in different computer vision tasks. Although harnessing the capability of Transformer-based architectures for medical image analysis seems interesting, these architectures are large, have a significant number of trainable parameters, and require large datasets to be trained on, which are usually rare in the medical domain. It has been claimed and empirically proved that at least part of the superior performance of Transformer-based architectures in Computer Vision domain originates from patch embedding operation. In this paper, we borrowed the previously introduced idea of integrating a fully Convolutional Neural Network architecture with Patch Embedding operation and presented an efficient CNN architecture for breast cancer malignancy detection from histopathological images. Despite the number of parameters that is significantly smaller than other methods, the accuracy performance metrics achieved 97.65%, 98.92%, 99.21%, and 98.01% for 40x, 100x, 200x, and 400x magnifications respectively. We took a step forward and modified the architecture using Group Convolution and Channel Shuffling ideas and reduced the number of trainable parameters even more with a negligible decline in performance and achieved 95.42%, 98.16%, 96.05%, and 97.92% accuracy for the mentioned magnifications respectively.

We investigate pre-training techniques for abstractive multi-document summarization (MDS), which is much less studied than summarizing single documents. Though recent work has demonstrated the effectiveness of highlighting information salience for pre-training strategy design, it struggles to generate abstractive and reflective summaries, which are critical properties for MDS. To this end, we present PELMS, a pre-trained model that uses objectives based on semantic coherence heuristics and faithfulness constraints with un-labeled multi-document inputs, to promote the generation of concise, fluent, and faithful summaries. To support the training of PELMS, we compile MultiPT, a multi-document pre-training corpus containing over 93 million documents to form more than 3 million unlabeled topic-centric document clusters, covering diverse genres such as product reviews, news, and general knowledge. We perform extensive evaluation of PELMS in low-shot settings on a wide range of MDS datasets. Our approach consistently outperforms competitive comparisons with respect to overall informativeness, abstractiveness, coherence, and faithfulness.

Automatic evaluation of generated textual content presents an ongoing challenge within the field of NLP. Given the impressive capabilities of modern language models (LMs) across diverse NLP tasks, there is a growing trend to employ these models in creating innovative evaluation metrics for automated assessment of generation tasks. This paper investigates a pivotal question: Do language model-driven evaluation metrics inherently exhibit bias favoring texts generated by the same underlying language model? Specifically, we assess whether prominent LM-based evaluation metrics--namely, BARTScore, T5Score, and GPTScore--demonstrate a favorable bias toward their respective underlying LMs in the context of summarization tasks. Our findings unveil a latent bias, particularly pronounced when such evaluation metrics are used in an reference-free manner without leveraging gold summaries. These results underscore that assessments provided by generative evaluation models can be influenced by factors beyond the inherent text quality, highlighting the necessity of developing more dependable evaluation protocols in the future.

Prevailing research practice today often relies on training dense retrievers on existing large datasets such as MSMARCO and then experimenting with ways to improve zero-shot generalization capabilities to unseen domains. While prior work has tackled this challenge through resource-intensive steps such as data augmentation, architectural modifications, increasing model size, or even further base model pretraining, comparatively little investigation has examined whether the training procedures themselves can be improved to yield better generalization capabilities in the resulting models. In this work, we recommend a simple recipe for training dense encoders: Train on MSMARCO with parameter-efficient methods, such as LoRA, and opt for using in-batch negatives unless given well-constructed hard negatives. We validate these recommendations using the BEIR benchmark and find results are persistent across choice of dense encoder and base model size and are complementary to other resource-intensive strategies for out-of-domain generalization such as architectural modifications or additional pretraining. We hope that this thorough and impartial study around various training techniques, which augments other resource-intensive methods, offers practical insights for developing a dense retrieval model that effectively generalizes, even when trained on a single dataset. Dense neural retrieval methods have been proven to be generally effective in many Information Retrieval (IR) tasks (Karpukhin et al., 2020; Izacard et al., 2021; Ni et al., 2021a). These methods use learned neural encoders to obtain dense vector representations of text and the relevance of passages for any given query is estimated by computing the dot product between their encodings. Dense approaches can outperform traditional retrieval techniques (e.g., BM25 (Robertson & Jones, 1976)), as they estimate similarity beyond syntactic matching (Lin et al., 2022). Neural retrieval models are effective rankers in domains for which large supervised datasets exist (e.g., MSMARCO (Campos et al., 2016) or Google NQ (Kwiatkowski et al., 2019)). Conversely, they might struggle to generalize to settings they have not been trained on, leading to challenges in handling out-ofdomain tasks (Thakur et al., 2021a; Ren et al., 2022; Lupart et al., 2023). In most real-world applications, supervision data is not available; whereas, retrieval models play a key role in the nascent field of augmented language models across many new exciting scenarios (Mialon et al., 2023).

This study introduces a new long-form database question answering dataset designed to evaluate how Large Language Models (LLMs) interact with a SQL interpreter. The task necessitates LLMs to strategically generate multiple SQL queries to retrieve sufficient data from a database, to reason with the acquired context, and to synthesize them into a comprehensive analytical narrative. Our findings highlight that this task poses great challenges even for the state-of-the-art GPT-4 model. We propose and evaluate two interaction strategies, and provide a fine-grained analysis of the individual stages within the interaction. A key discovery is the identification of two primary bottlenecks hindering effective interaction: the capacity for planning and the ability to generate multiple SQL queries. To address the challenge of accurately assessing answer quality, we introduce a multi-agent evaluation framework that simulates the academic peer-review process, enhancing the precision and reliability of our evaluations. This framework allows for a more nuanced understanding of the strengths and limitations of current LLMs in complex retrieval and reasoning tasks.

Are the longstanding robustness issues in NLP resolved by today's larger and more performant models? To address this question, we conduct a thorough investigation using 19 models of different sizes spanning different architectural choices and pretraining objectives. We conduct evaluations using (a) OOD and challenge test sets, (b) CheckLists, (c) contrast sets, and (d) adversarial inputs. Our analysis reveals that not all OOD tests provide further insight into robustness. Evaluating with CheckLists and contrast sets shows significant gaps in model performance; merely scaling models does not make them sufficiently robust. Finally, we point out that current approaches for adversarial evaluations of models are themselves problematic: they can be easily thwarted, and in their current forms, do not represent a sufficiently deep probe of model robustness. We conclude that not only is the question of robustness in NLP as yet unresolved, but even some of the approaches to measure robustness need to be reassessed.

Large language models (LLMs) have the ability to perform in-context learning (ICL) of new tasks by conditioning on prompts comprising a few task examples. This work studies the problem of selecting the best examples given a candidate pool to improve ICL performance on given a test input. Existing approaches either require training with feedback from a much larger LLM or are computationally expensive. We propose a novel metric, GistScore, based on Example Gisting, a novel approach for training example retrievers for ICL using an attention bottleneck via Gisting, a recent technique for compressing task instructions. To tradeoff performance with ease of use, we experiment with both fine-tuning gist models on each dataset and multi-task training a single model on a large collection of datasets. On 21 diverse datasets spanning 9 tasks, we show that our fine-tuned models get state-of-the-art ICL performance with 20% absolute average gain over off-the-shelf retrievers and 7% over the best prior methods. Our multi-task model generalizes well out-of-the-box to new task categories, datasets, and prompt templates with retrieval speeds that are consistently thousands of times faster than the best prior training-free method.

Ensuring a neural network is not relying on protected attributes (e.g., race, sex, age) for prediction is crucial in advancing fair and trustworthy AI. While several promising methods for removing attribute bias in neural networks have been proposed, their limitations remain under-explored. To that end, in this work, we mathematically and empirically reveal the limitation of existing attribute bias removal methods in presence of strong bias and propose a new method that can mitigate this limitation. Specifically, we first derive a general non-vacuous information-theoretical upper bound on the performance of any attribute bias removal method in terms of the bias strength, revealing that they are effective only when the inherent bias in the dataset is relatively weak. Next, we derive a necessary condition for the existence of any method that can remove attribute bias regardless of the bias strength. Inspired by this condition, we then propose a new method using an adversarial objective that directly filters out protected attributes in the input space while maximally preserving all other attributes, without requiring any specific target label. The proposed method achieves state-of-the-art performance in both strong and moderate bias settings. We provide extensive experiments on synthetic, image, and census datasets, to verify the derived theoretical bound and its consequences in practice, and evaluate the effectiveness of the proposed method in removing strong attribute bias.

Large language models (LLMs) demonstrate emergent in-context learning capabilities, where they adapt to new tasks based on example demonstrations. However, in-context learning has seen limited effectiveness in many settings, is difficult to quantitatively control and takes up context window space. To overcome these limitations, we propose an alternative approach that recasts in-context learning as in-context vectors (ICV). Using ICV has two steps. We first use a forward pass on demonstration examples to create the in-context vector from the latent embedding of the LLM. This vector captures essential information about the intended task. On a new query, instead of adding demonstrations to the prompt, we shift the latent states of the LLM using the ICV. The ICV approach has several benefits: 1) it enables the LLM to more effectively follow the demonstration examples; 2) it's easy to control by adjusting the magnitude of the ICV; 3) it reduces the length of the prompt by removing the in-context demonstrations; 4) ICV is computationally much more efficient than fine-tuning. We demonstrate that ICV achieves better performance compared to standard in-context learning and fine-tuning on diverse tasks including safety, style transfer, role-playing and formatting. Moreover, we show that we can flexibly teach LLM to simultaneously follow different types of instructions by simple vector arithmetics on the corresponding ICVs.