To solve complex tasks, large language models (LLMs) often require multiple rounds of interactions with the user, sometimes assisted by external tools. However, current evaluation paradigms often focus solely on benchmark performance with single-turn exchanges, neglecting the intricate interactions among the user, LLMs, and external tools, creating a discrepancy between benchmark evaluation and real-world use cases. We introduce MINT benchmark to evaluate LLMs' ability to solve tasks with multi-turn interactions by (1) using tools and (2) leveraging natural language feedback. To ensure reproducibility, we provide an evaluation framework where LLMs can access tools by executing Python code and receive natural language feedback from the user simulated with GPT-4. We repurpose a diverse set of established datasets and tasks focusing on reasoning, coding, and decision-making and carefully curate them into a compact subset of instances for efficient evaluation. Our analysis of 20 open- and closed-source LLMs offers intriguing findings. (1) LLMs generally benefit from tool interactions and language feedback, with performance gains (absolute, same below) of 1--8% per additional turn with tool use and 2--17% with natural language feedback. (2) Better single-turn performance does not guarantee better multi-turn performance. (3) Surprisingly, on LLMs we evaluated, we found supervised instruction-finetuning (SIFT) and reinforcement learning from human feedback (RLHF) generally hurt multi-turn capabilities. We hope MINT can help measure progress and incentivize research in improving LLMs' capabilities in multi-turn interactions, especially for open-source communities where multi-turn human evaluation has been less accessible compared to commercial LLMs with a larger user base.
Skin lesion segmentation is a fundamental task in dermoscopic image analysis. The complex features of pixels in the lesion region impede the lesion segmentation accuracy, and existing deep learning-based methods often lack interpretability to this problem. In this work, we propose a novel unsupervised Skin Lesion sEgmentation framework based on structural entropy and isolation forest outlier Detection, namely SLED. Specifically, skin lesions are segmented by minimizing the structural entropy of a superpixel graph constructed from the dermoscopic image. Then, we characterize the consistency of healthy skin features and devise a novel multi-scale segmentation mechanism by outlier detection, which enhances the segmentation accuracy by leveraging the superpixel features from multiple scales. We conduct experiments on four skin lesion benchmarks and compare SLED with nine representative unsupervised segmentation methods. Experimental results demonstrate the superiority of the proposed framework. Additionally, some case studies are analyzed to demonstrate the effectiveness of SLED.
With the proliferation of social media, a growing number of users search for and join group activities in their daily life. This develops a need for the study on the group identification (GI) task, i.e., recommending groups to users. The major challenge in this task is how to predict users' preferences for groups based on not only previous group participation of users but also users' interests in items. Although recent developments in Graph Neural Networks (GNNs) accomplish embedding multiple types of objects in graph-based recommender systems, they, however, fail to address this GI problem comprehensively. In this paper, we propose a novel framework named Group Identification via Transitional Hypergraph Convolution with Graph Self-supervised Learning (GTGS). We devise a novel transitional hypergraph convolution layer to leverage users' preferences for items as prior knowledge when seeking their group preferences. To construct comprehensive user/group representations for GI task, we design the cross-view self-supervised learning to encourage the intrinsic consistency between item and group preferences for each user, and the group-based regularization to enhance the distinction among group embeddings. Experimental results on three benchmark datasets verify the superiority of GTGS. Additional detailed investigations are conducted to demonstrate the effectiveness of the proposed framework.
Rising computational demands of modern natural language processing (NLP) systems have increased the barrier to entry for cutting-edge research while posing serious environmental concerns. Yet, progress on model efficiency has been impeded by practical challenges in model evaluation and comparison. For example, hardware is challenging to control due to disparate levels of accessibility across different institutions. Moreover, improvements in metrics such as FLOPs often fail to translate to progress in real-world applications. In response, we introduce Pentathlon, a benchmark for holistic and realistic evaluation of model efficiency. Pentathlon focuses on inference, which accounts for a majority of the compute in a model's lifecycle. It offers a strictly-controlled hardware platform, and is designed to mirror real-world applications scenarios. It incorporates a suite of metrics that target different aspects of efficiency, including latency, throughput, memory overhead, and energy consumption. Pentathlon also comes with a software library that can be seamlessly integrated into any codebase and enable evaluation. As a standardized and centralized evaluation platform, Pentathlon can drastically reduce the workload to make fair and reproducible efficiency comparisons. While initially focused on natural language processing (NLP) models, Pentathlon is designed to allow flexible extension to other fields. We envision Pentathlon will stimulate algorithmic innovations in building efficient models, and foster an increased awareness of the social and environmental implications in the development of future-generation NLP models.
Recommendation systems suffer in the strict cold-start (SCS) scenario, where the user-item interactions are entirely unavailable. The ID-based approaches completely fail to work. Cold-start recommenders, on the other hand, leverage item contents to map the new items to the existing ones. However, the existing SCS recommenders explore item contents in coarse-grained manners that introduce noise or information loss. Moreover, informative data sources other than item contents, such as users' purchase sequences and review texts, are ignored. We explore the role of the fine-grained item attributes in bridging the gaps between the existing and the SCS items and pre-train a knowledgeable item-attribute graph for SCS item recommendation. Our proposed framework, ColdGPT, models item-attribute correlations into an item-attribute graph by extracting fine-grained attributes from item contents. ColdGPT then transfers knowledge into the item-attribute graph from various available data sources, i.e., item contents, historical purchase sequences, and review texts of the existing items, via multi-task learning. To facilitate the positive transfer, ColdGPT designs submodules according to the natural forms of the data sources and coordinates the multiple pre-training tasks via unified alignment-and-uniformity losses. Our pre-trained item-attribute graph acts as an implicit, extendable item embedding matrix, which enables the SCS item embeddings to be easily acquired by inserting these items and propagating their attributes' embeddings. We carefully process three public datasets, i.e., Yelp, Amazon-home, and Amazon-sports, to guarantee the SCS setting for evaluation. Extensive experiments show that ColdGPT consistently outperforms the existing SCS recommenders by large margins and even surpasses models that are pre-trained on 75-224 times more, cross-domain data on two out of four datasets.
Sequential recommendation (SR) investigates the dynamic user preferences modeling and generates the next-item prediction. The next item preference is typically generated by the affinity between the sequence and item representations. However, both sequence and item representations suffer from the rank degeneration issue due to the data sparsity problem. The rank degeneration issue significantly impairs the representations for SR. This motivates us to measure how severe is the rank degeneration issue and alleviate the sequence and item representation rank degeneration issues simultaneously for SR. In this work, we theoretically connect the sequence representation degeneration issue with the item rank degeneration, particularly for short sequences and cold items. We also identify the connection between the fast singular value decay phenomenon and the rank collapse issue in transformer sequence output and item embeddings. We propose the area under the singular value curve metric to evaluate the severity of the singular value decay phenomenon and use it as an indicator of rank degeneration. We further introduce a novel singular spectrum smoothing regularization to alleviate the rank degeneration on both sequence and item sides, which is the Singular sPectrum sMoothing for sequential Recommendation (SPMRec). We also establish a correlation between the ranks of sequence and item embeddings and the rank of the user-item preference prediction matrix, which can affect recommendation diversity. We conduct experiments on four benchmark datasets to demonstrate the superiority of SPMRec over the state-of-the-art recommendation methods, especially in short sequences. The experiments also demonstrate a strong connection between our proposed singular spectrum smoothing and recommendation diversity.
The unprecedented performance of large language models (LLMs) necessitates improvements in evaluations. Rather than merely exploring the breadth of LLM abilities, we believe meticulous and thoughtful designs are essential to thorough, unbiased, and applicable evaluations. Given the importance of world knowledge to LLMs, we construct a Knowledge-oriented LLM Assessment benchmark (KoLA), in which we carefully design three crucial factors: (1) For ability modeling, we mimic human cognition to form a four-level taxonomy of knowledge-related abilities, covering $19$ tasks. (2) For data, to ensure fair comparisons, we use both Wikipedia, a corpus prevalently pre-trained by LLMs, along with continuously collected emerging corpora, aiming to evaluate the capacity to handle unseen data and evolving knowledge. (3) For evaluation criteria, we adopt a contrastive system, including overall standard scores for better numerical comparability across tasks and models and a unique self-contrast metric for automatically evaluating knowledge hallucination. We evaluate $21$ open-source and commercial LLMs and obtain some intriguing findings. The KoLA dataset and open-participation leaderboard are publicly released at https://kola.xlore.cn and will be continuously updated to provide references for developing LLMs and knowledge-related systems.
Event extraction (EE) is a crucial task aiming at extracting events from texts, which includes two subtasks: event detection (ED) and event argument extraction (EAE). In this paper, we check the reliability of EE evaluations and identify three major pitfalls: (1) The data preprocessing discrepancy makes the evaluation results on the same dataset not directly comparable, but the data preprocessing details are not widely noted and specified in papers. (2) The output space discrepancy of different model paradigms makes different-paradigm EE models lack grounds for comparison and also leads to unclear mapping issues between predictions and annotations. (3) The absence of pipeline evaluation of many EAE-only works makes them hard to be directly compared with EE works and may not well reflect the model performance in real-world pipeline scenarios. We demonstrate the significant influence of these pitfalls through comprehensive meta-analyses of recent papers and empirical experiments. To avoid these pitfalls, we suggest a series of remedies, including specifying data preprocessing, standardizing outputs, and providing pipeline evaluation results. To help implement these remedies, we develop a consistent evaluation framework OMNIEVENT, which can be obtained from https://github.com/THU-KEG/OmniEvent.
As large language models (LLMs) are continuously being developed, their evaluation becomes increasingly important yet challenging. This work proposes Chain-of-Thought Hub, an open-source evaluation suite on the multi-step reasoning capabilities of large language models. We are interested in this setting for two reasons: (1) from the behavior of GPT and PaLM model family, we observe that complex reasoning is likely to be a key differentiator between weaker and stronger LLMs; (2) we envisage large language models to become the next-generation computational platform and foster an ecosystem of LLM-based new applications, this naturally requires the foundation models to perform complex tasks that often involve the composition of linguistic and logical operations. Our approach is to compile a suite of challenging reasoning benchmarks to track the progress of LLMs. Our current results show that: (1) model scale clearly correlates with reasoning capabilities; (2) As of May 2023, Claude-v1.3 and PaLM-2 are the only two models that are comparable with GPT-4, while open-sourced models still lag behind; (3) LLaMA-65B performs closely to code-davinci-002, indicating that with successful further development such as reinforcement learning from human feedback (RLHF), it has great potential to be close to GPT-3.5-Turbo. Our results also suggest that for the open-source efforts to catch up, the community may focus more on building better base models and exploring RLHF.
Finetuning pre-trained language models (LMs) enhances the models' capabilities. Prior techniques fine-tune a pre-trained LM on input-output pairs (e.g., instruction fine-tuning), or with numerical rewards that gauge the quality of its outputs (e.g., reinforcement learning from human feedback). We explore LMs' potential to learn from textual interactions (LeTI) that not only check their correctness with binary labels, but also pinpoint and explain errors in their outputs through textual feedback. Our investigation focuses on the code generation task, where the model produces code pieces in response to natural language instructions. This setting invites a natural and scalable way to acquire the textual feedback: the error messages and stack traces from code execution using a Python interpreter. LeTI iteratively fine-tunes the model, using the LM objective, on a concatenation of natural language instructions, LM-generated programs, and textual feedback, which is only provided when the generated program fails to solve the task. Prepended to this fine-tuning text, a binary reward token is used to differentiate correct and buggy solutions. On MBPP, a code generation dataset, LeTI substantially improves the performance of two base LMs of different scales. LeTI requires no ground-truth outputs for training and even outperforms a fine-tuned baseline that does. LeTI's strong performance generalizes to other datasets. Trained on MBPP, it achieves comparable or better performance than the base LMs on unseen problems in HumanEval. Furthermore, compared to binary feedback, we observe that textual feedback leads to improved generation quality and sample efficiency, achieving the same performance with fewer than half of the gradient steps. LeTI is equally applicable in natural language tasks when they can be formulated as code generation, which we empirically verified on event argument extraction.