Overview of the CAIL 2023 Argument Mining Track

We give a detailed overview of the CAIL 2023 Argument Mining Track, one of the Chinese AI and Law Challenge (CAIL) 2023 tracks. The main goal of the track is to identify and extract interacting argument pairs in trial dialogs. It mainly uses summarized judgment documents but can also refer to trial recordings. The track consists of two stages, and we introduce the tasks designed for each stage; we also extend the data from previous events into a new dataset -- CAIL2023-ArgMine -- with annotated new cases from various causes of action. We outline several submissions that achieve the best results, including their methods for different stages. While all submissions rely on language models, they have incorporated strategies that may benefit future work in this field.

What Kinds of Tokens Benefit from Distant Text? An Analysis on Long Context Language Modeling

As the context length that large language models can handle continues to increase, these models demonstrate an enhanced ability to utilize distant information for tasks such as language modeling. This capability contrasts with human reading and writing habits, where it is uncommon to remember and use particularly distant information, except in cases of foreshadowing. In this paper, we aim to explore which kinds of words benefit more from long contexts in language models. By analyzing the changes in token probabilities with increasing context length, we find that content words (e.g., nouns, adjectives) and the initial tokens of words benefit the most. Frequent patterns in the context (N-grams) also significantly impact predictions. Additionally, the model's prior knowledge plays a crucial role in influencing predictions, especially for rare tokens. We also observe that language models become more confident with longer contexts, resulting in sharper probability distributions. This overconfidence may contribute to the increasing probabilities of tokens with distant contextual information. We hope that our analysis will help the community better understand long-text language modeling and contribute to the design of more reliable long-context models.

DiNeR: a Large Realistic Dataset for Evaluating Compositional Generalization

Most of the existing compositional generalization datasets are synthetically-generated, resulting in a lack of natural language variation. While there have been recent attempts to introduce non-synthetic datasets for compositional generalization, they suffer from either limited data scale or a lack of diversity in the forms of combinations. To better investigate compositional generalization with more linguistic phenomena and compositional diversity, we propose the DIsh NamE Recognition (DiNeR) task and create a large realistic Chinese dataset. Given a recipe instruction, models are required to recognize the dish name composed of diverse combinations of food, actions, and flavors. Our dataset consists of 3,811 dishes and 228,114 recipes, and involves plenty of linguistic phenomena such as anaphora, omission and ambiguity. We provide two strong baselines based on T5 and large language models (LLMs). This work contributes a challenging task, baseline methods to tackle the task, and insights into compositional generalization in the context of dish name recognition. Code and data are available at https://github.com/Jumpy-pku/DiNeR.

Can Perplexity Reflect Large Language Model's Ability in Long Text Understanding?

Recent studies have shown that Large Language Models (LLMs) have the potential to process extremely long text. Many works only evaluate LLMs' long-text processing ability on the language modeling task, with perplexity (PPL) as the evaluation metric. However, in our study, we find that there is no correlation between PPL and LLMs' long-text understanding ability. Besides, PPL may only reflect the model's ability to model local information instead of catching long-range dependency. Therefore, only using PPL to prove the model could process long text is inappropriate. The local focus feature of PPL could also explain some existing phenomena, such as the great extrapolation ability of the position method ALiBi. When evaluating a model's ability in long text, we might pay more attention to PPL's limitation and avoid overly relying on it.

Motion Generation from Fine-grained Textual Descriptions

The task of text2motion is to generate human motion sequences from given textual descriptions, where the model explores diverse mappings from natural language instructions to human body movements. While most existing works are confined to coarse-grained motion descriptions, e.g., "A man squats.", fine-grained descriptions specifying movements of relevant body parts are barely explored. Models trained with coarse-grained texts may not be able to learn mappings from fine-grained motion-related words to motion primitives, resulting in the failure to generate motions from unseen descriptions. In this paper, we build a large-scale language-motion dataset specializing in fine-grained textual descriptions, FineHumanML3D, by feeding GPT-3.5-turbo with step-by-step instructions with pseudo-code compulsory checks. Accordingly, we design a new text2motion model, FineMotionDiffuse, making full use of fine-grained textual information. Our quantitative evaluation shows that FineMotionDiffuse trained on FineHumanML3D improves FID by a large margin of 0.38, compared with competitive baselines. According to the qualitative evaluation and case study, our model outperforms MotionDiffuse in generating spatially or chronologically composite motions, by learning the implicit mappings from fine-grained descriptions to the corresponding basic motions. We release our data at https://github.com/KunhangL/finemotiondiffuse.

Harder Tasks Need More Experts: Dynamic Routing in MoE Models

In this paper, we introduce a novel dynamic expert selection framework for Mixture of Experts (MoE) models, aiming to enhance computational efficiency and model performance by adjusting the number of activated experts based on input difficulty. Unlike traditional MoE approaches that rely on fixed Top-K routing, which activates a predetermined number of experts regardless of the input's complexity, our method dynamically selects experts based on the confidence level in expert selection for each input. This allows for a more efficient utilization of computational resources, activating more experts for complex tasks requiring advanced reasoning and fewer for simpler tasks. Through extensive evaluations, our dynamic routing method demonstrates substantial improvements over conventional Top-2 routing across various benchmarks, achieving an average improvement of 0.7% with less than 90% activated parameters. Further analysis shows our model dispatches more experts to tasks requiring complex reasoning skills, like BBH, confirming its ability to dynamically allocate computational resources in alignment with the input's complexity. Our findings also highlight a variation in the number of experts needed across different layers of the transformer model, offering insights into the potential for designing heterogeneous MoE frameworks. The code and models are available at https://github.com/ZhenweiAn/Dynamic_MoE.

ProTrix: Building Models for Planning and Reasoning over Tables with Sentence Context

Tables play a crucial role in conveying information in various domains, serving as indispensable tools for organizing and presenting data in a structured manner. We propose a Plan-then-Reason framework to answer different types of user queries over tables with sentence context. The framework first plans the reasoning paths over the context, then assigns each step to program-based or textual reasoning to reach the final answer. We construct an instruction tuning set TrixInstruct following the framework. Our dataset cover queries that are program-unsolvable or need combining information from tables and sentences to obtain planning and reasoning abilities. We present ProTrix by finetuning Llama-2-7B on TrixInstruct. Our experiments show that ProTrix generalizes to diverse tabular tasks and achieves comparable performance to GPT-3.5-turbo. We further demonstrate that ProTrix can generate accurate and faithful explanations to answer complex free-form questions. Our work underscores the importance of the planning and reasoning abilities towards a model over tabular tasks with generalizability and interpretability. We will release our dataset and model at https://github.com/WilliamZR/ProTrix.

Teaching Large Language Models an Unseen Language on the Fly

Existing large language models struggle to support numerous low-resource languages, particularly the extremely low-resource ones where there is minimal training data available for effective parameter updating. We thus investigate whether LLMs can learn a new language on the fly solely through prompting. To study this question, we collect a research suite for Zhuang, a language supported by no LLMs currently. We introduce \textsc{DiPMT++}, a framework for adapting LLMs to unseen languages by in-context learning. Using a dictionary and only 5K parallel sentences, \textsc{DiPMT++} significantly enhances the performance of GPT-4 from 0 to 16 BLEU for Chinese-to-Zhuang translation and achieves 32 BLEU for Zhuang-to-Chinese translation. Furthermore, we demonstrate the practical utility of this framework in aiding humans to translate completely unseen languages, which could contribute to the preservation of linguistic diversity.

Probing Multimodal Large Language Models for Global and Local Semantic Representation

The success of large language models has inspired researchers to transfer their exceptional representing ability to other modalities. Several recent works leverage image-caption alignment datasets to train multimodal large language models (MLLMs), which achieve state-of-the-art performance on image-to-text tasks. However, there are very few studies exploring whether MLLMs truly understand the complete image information, i.e., global information, or if they can only capture some local object information. In this study, we find that the intermediate layers of models can encode more global semantic information, whose representation vectors perform better on visual-language entailment tasks, rather than the topmost layers. We further probe models for local semantic representation through object detection tasks. And we draw a conclusion that the topmost layers may excessively focus on local information, leading to a diminished ability to encode global information.

Are LLMs Capable of Data-based Statistical and Causal Reasoning? Benchmarking Advanced Quantitative Reasoning with Data

Quantitative reasoning is a critical skill to analyze data, yet the assessment of such ability remains limited. To address this gap, we introduce the Quantitative Reasoning with Data (QRData) benchmark, aiming to evaluate Large Language Models' capability in statistical and causal reasoning with real-world data. The benchmark comprises a carefully constructed dataset of 411 questions accompanied by data sheets from textbooks, online learning materials, and academic papers. To compare models' quantitative reasoning abilities on data and text, we enrich the benchmark with an auxiliary set of 290 text-only questions, namely QRText. We evaluate natural language reasoning, program-based reasoning, and agent reasoning methods including Chain-of-Thought, Program-of-Thoughts, ReAct, and code interpreter assistants on diverse models. The strongest model GPT-4 achieves an accuracy of 58%, which has a large room for improvement. Among open-source models, Deepseek-coder-instruct, a code LLM pretrained on 2T tokens, gets the highest accuracy of 37%. Analysis reveals that models encounter difficulties in data analysis and causal reasoning, and struggle in using causal knowledge and provided data simultaneously. Code and data are in https://github.com/xxxiaol/QRData.