Deep learning models often exhibit overconfidence in predicting
out-of-distribution (OOD) data, underscoring the crucial role of OOD detection
in ensuring reliability in predictions. Among various OOD detection approaches,
post-hoc detectors have gained significant popularity, primarily due to their
ease of use and implementation. However, the effectiveness of most post-hoc OOD
detectors has been constrained as they rely solely either on extreme
information, such as the maximum logit, or on the collective information (i.e.,
information spanned across classes or training samples) embedded within the
output layer. In this paper, we propose ExCeL that combines both extreme and
collective information within the output layer for enhanced accuracy in OOD
detection. We leverage the logit of the top predicted class as the extreme
information (i.e., the maximum logit), while the collective information is
derived in a novel approach that involves assessing the likelihood of other
classes appearing in subsequent ranks across various training samples. Our idea
is motivated by the observation that, for in-distribution (ID) data, the
ranking of classes beyond the predicted class is more deterministic compared to
that in OOD data. Experiments conducted on CIFAR100 and ImageNet-200 datasets
demonstrate that ExCeL consistently is among the five top-performing methods
out of twenty-one existing post-hoc baselines when the joint performance on
near-OOD and far-OOD is considered (i.e., in terms of AUROC and FPR95).
Furthermore, ExCeL shows the best overall performance across both datasets,
unlike other baselines that work best on one dataset but has a performance drop
in the other.

使用 ExCeL 模型结合极端信息和集合信息，提高了在预测时对于认知外数据（OOD）检测的准确性，并在 CIFAR100 和 ImageNet-200 数据集上实现了最佳性能。

ExCeL : 组合极值与集体逻辑信息以增强离群检测

ExCeL : Combined Extreme and Collective Logit Information for Enhancing  Out-of-Distribution Detection

With the evolution of Large Language Models (LLMs) we can solve increasingly
more complex NLP tasks across various domains, including spreadsheets. This
work investigates whether LLMs can generate code (Excel OfficeScripts, a
TypeScript API for executing many tasks in Excel) that solves Excel specific
tasks provided via natural language user instructions. To do so we introduce a
new large-scale benchmark, InstructExcel, created by leveraging the 'Automate'
feature in Excel to automatically generate OfficeScripts from users' actions.
Our benchmark includes over 10k samples covering 170+ Excel operations across
2,000 publicly available Excel spreadsheets. Experiments across various
zero-shot and few-shot settings show that InstructExcel is a hard benchmark for
state of the art models like GPT-4. We observe that (1) using GPT-4 over
GPT-3.5, (2) providing more in-context examples, and (3) dynamic prompting can
help improve performance on this benchmark.

该研究探讨了大型语言模型在解决 Excel 相关任务上的可行性，并引入了一个新的大规模基准测试，InstructExcel，该测试覆盖了多个 Excel 操作并包含超过 10,000 个样本。实验结果表明，GPT-4 相比 GPT-3.5 在此基准测试上表现较好，并且提供更多上下文示例和动态提示可以提高性能。

InstructExcel：一个 Excel 自然语言指令的基准

InstructExcel: A Benchmark for Natural Language Instruction in Excel

Spreadsheet computing is one of the more popular computing methodologies in
today's modern society. The spreadsheet application's ease of use and
usefulness has enabled non-programmers to perform programming-like tasks in a
familiar setting modeled after the tabular "pen and paper" approach. However,
spreadsheet applications are limited to bookkeeping-like tasks due to their
single-direction data flow. This thesis demonstrates an extension of the
spreadsheet computing paradigm in overcoming this limitation to solve
constraint satisfaction problems. We present a framework seamlessly
incorporating a finite constraint solver with the spreadsheet computing
paradigm. This framework allows the individual cells in the spreadsheet to be
attached to either a finite domain or a constraint specifying the relationship
among the cells. The framework provides an interface for constraint solving and
further enhances the spreadsheet computing paradigm by providing a set of
spreadsheet-specific constraints that will aid in controlling the scalability
of large spreadsheet applications implementations. Finally, we provide examples
to demonstrate the usability and usefulness of the extended spreadsheet
paradigm.
Keywords: Spreadsheet computing, Constraint Logic Programming, Constraint
satisfaction, Domain-Specific language, Excel, SWI Prolog, C#

本文介绍了如何通过将有限约束求解器与电子表格计算范例结合来扩展电子表格计算范例应用，介绍了一个框架，将电子表格中的每个单元格与有限域或约束关系相连。同时提供了特定于电子表格的约束，以帮助控制大型电子表格应用程序实现的可扩展性，并通过示例演示了扩展的电子表格范例的可用性和有用性。