CODIS: Benchmarking Context-Dependent Visual Comprehension for Multimodal Large Language Models

Multimodal large language models (MLLMs) have demonstrated promising results in a variety of tasks that combine vision and language. As these models become more integral to research and applications, conducting comprehensive evaluations of their capabilities has grown increasingly important. However, most existing benchmarks fail to consider that, in certain situations, images need to be interpreted within a broader context. In this work, we introduce a new benchmark, named as CODIS, designed to assess the ability of models to use context provided in free-form text to enhance visual comprehension. Our findings indicate that MLLMs consistently fall short of human performance on this benchmark. Further analysis confirms that these models struggle to effectively extract and utilize contextual information to improve their understanding of images. This underscores the pressing need to enhance the ability of MLLMs to comprehend visuals in a context-dependent manner. View our project website at https://thunlp-mt.github.io/CODIS.

PANDA: Preference Adaptation for Enhancing Domain-Specific Abilities of LLMs

While Large language models (LLMs) have demonstrated considerable capabilities across various natural language tasks, they often fall short of the performance achieved by domain-specific state-of-the-art models. One potential approach to enhance domain-specific capabilities of LLMs involves fine-tuning them using corresponding datasets. However, this method can be both resource and time-intensive, and not applicable to closed-source commercial LLMs. In this paper, we propose Preference Adaptation for Enhancing Domain-specific Abilities of LLMs (PANDA), a method designed to augment the domain-specific capabilities of LLMs by leveraging insights from the response preference of expert models without requiring fine-tuning. Our experimental results reveal that PANDA significantly enhances the domain-specific ability of LLMs on text classification and interactive decision tasks. Moreover, LLM with PANDA even outperforms the expert model that being learned on 4 tasks of ScienceWorld. This finding highlights the potential of exploring tuning-free approaches to achieve weak-to-strong generalization.

Model Composition for Multimodal Large Language Models

Recent developments in Multimodal Large Language Models (MLLMs) have shown rapid progress, moving towards the goal of creating versatile MLLMs that understand inputs from various modalities. However, existing methods typically rely on joint training with paired multimodal instruction data, which is resource-intensive and challenging to extend to new modalities. In this paper, we propose a new paradigm through the model composition of existing MLLMs to create a new model that retains the modal understanding capabilities of each original model. Our basic implementation, NaiveMC, demonstrates the effectiveness of this paradigm by reusing modality encoders and merging LLM parameters. Furthermore, we introduce DAMC to address parameter interference and mismatch issues during the merging process, thereby enhancing the model performance. To facilitate research in this area, we propose MCUB, a benchmark for assessing ability of MLLMs to understand inputs from diverse modalities. Experiments on this benchmark and four other multimodal understanding tasks show significant improvements over baselines, proving that model composition can create a versatile model capable of processing inputs from multiple modalities.

Scaffolding Coordinates to Promote Vision-Language Coordination in Large Multi-Modal Models

State-of-the-art Large Multi-Modal Models (LMMs) have demonstrated exceptional capabilities in vision-language tasks. Despite their advanced functionalities, the performances of LMMs are still limited in challenging scenarios that require complex reasoning with multiple levels of visual information. Existing prompting techniques for LMMs focus on either improving textual reasoning or leveraging tools for image preprocessing, lacking a simple and general visual prompting scheme to promote vision-language coordination in LMMs. In this work, we propose Scaffold prompting that scaffolds coordinates to promote vision-language coordination. Specifically, Scaffold overlays a dot matrix within the image as visual information anchors and leverages multi-dimensional coordinates as textual positional references. Extensive experiments on a wide range of challenging vision-language tasks demonstrate the superiority of Scaffold over GPT-4V with the textual CoT prompting. Our code is released in https://github.com/leixy20/Scaffold.

Enhancing Multilingual Capabilities of Large Language Models through Self-Distillation from Resource-Rich Languages

While large language models (LLMs) have been pre-trained on multilingual corpora, their performance still lags behind in most languages compared to a few resource-rich languages. One common approach to mitigate this issue is to translate training data from resource-rich languages into other languages and then continue training. However, using the data obtained solely relying on translation while ignoring the original capabilities of LLMs across languages is not always effective, which we show will limit the performance of cross-lingual knowledge transfer. In this work, we propose SDRRL, a method based on Self-Distillation from Resource-Rich Languages that effectively improve multilingual performance by leveraging the internal capabilities of LLMs on resource-rich languages. We evaluate on different LLMs (LLaMA-2 and SeaLLM) and source languages across various comprehension and generation tasks, experimental results demonstrate that SDRRL can significantly enhance multilingual capabilities while minimizing the impact on original performance in resource-rich languages.

Browse and Concentrate: Comprehending Multimodal Content via prior-LLM Context Fusion

With the bloom of Large Language Models (LLMs), Multimodal Large Language Models (MLLMs) that incorporate LLMs with pre-trained vision models have recently demonstrated impressive performance across diverse vision-language tasks. However, they fall short to comprehend context involving multiple images. A primary reason for this shortcoming is that the visual features for each images are encoded individually by frozen encoders before feeding into the LLM backbone, lacking awareness of other images and the multimodal instructions. We term this issue as prior-LLM modality isolation and propose a two phase paradigm, browse-and-concentrate, to enable in-depth multimodal context fusion prior to feeding the features into LLMs. This paradigm initially "browses" through the inputs for essential insights, and then revisits the inputs to "concentrate" on crucial details, guided by these insights, to achieve a more comprehensive understanding of the multimodal inputs. Additionally, we develop training strategies specifically to enhance the understanding of multi-image inputs. Our method markedly boosts the performance on 7 multi-image scenarios, contributing to increments on average accuracy by 2.13% and 7.60% against strong MLLMs baselines with 3B and 11B LLMs, respectively.

Meta Ranking: Less Capable Language Models are Capable for Single Response Judgement

Although Large Language Models (LLMs) have demonstrated strong performance on a wide range of tasks, they still face reliability challenges such as hallucination. Previous studies reveal that highly capable LLMs like GPT-4 are effective in judging the reliability of individual responses, while less capable ones are often tuned to evaluate the relative reliability of responses to the same query. To enable less capable LLMs to effectively judge the reliability of individual responses, we propose a novel method named $\textit{Meta}$ $\textit{Ranking}$ (MR). Unlike previous methods, which assess the response directly, we achieve the judgement by comparing the target query-response pair with reference query-response pairs. We found its remarkable effectiveness in error detection for LLM responses on reasoning tasks, where less capable LLMs could outperform strong baselines, even without fine-tuning. We further demonstrate that MR can be used to enhance the performance of LLMs in two practical applications: query routing and iterative training data filtering. The former achieves GPT-4-turbo comparable performance with less than half the token consumption, while the latter makes the instruction-tuned LLaMA-7B and Phi-2, a 2.7B model, significantly surpass Alpaca-13B over fewer training samples, underscoring the high potential of our proposed method.

Towards Unified Alignment Between Agents, Humans, and Environment

The rapid progress of foundation models has led to the prosperity of autonomous agents, which leverage the universal capabilities of foundation models to conduct reasoning, decision-making, and environmental interaction. However, the efficacy of agents remains limited when operating in intricate, realistic environments. In this work, we introduce the principles of $\mathbf{U}$nified $\mathbf{A}$lignment for $\mathbf{A}$gents ($\mathbf{UA}^2$), which advocate for the simultaneous alignment of agents with human intentions, environmental dynamics, and self-constraints such as the limitation of monetary budgets. From the perspective of $\mathbf{UA}^2$, we review the current agent research and highlight the neglected factors in existing agent benchmarks and method candidates. We also conduct proof-of-concept studies by introducing realistic features to WebShop, including user profiles to demonstrate intentions, personalized reranking for complex environmental dynamics, and runtime cost statistics to reflect self-constraints. We then follow the principles of $\mathbf{UA}^2$ to propose an initial design of our agent, and benchmark its performance with several candidate baselines in the retrofitted WebShop. The extensive experimental results further prove the importance of the principles of $\mathbf{UA}^2$. Our research sheds light on the next steps of autonomous agent research with improved general problem-solving abilities.

Speak It Out: Solving Symbol-Related Problems with Symbol-to-Language Conversion for Language Models

Symbols (or more broadly, non-natural language textual representations) such as numerical sequences, molecular formulas, and table delimiters widely exist, playing important roles in various tasks such as abstract reasoning, chemical property prediction, and table question answering. Despite the impressive natural language comprehension capabilities of large language models (LLMs), their reasoning abilities for symbols remain inadequate, which could attributed to the difference between symbol representations and general natural languages. We propose symbol-to-language (S2L), a tuning-free method that enables large language models to solve symbol-related problems with information expressed in natural language. Specifically, S2L first converts the symbols involved to language-based representations, which can be implemented by prompting LLMs or leveraging external tools, then these language-based representations are integrated into the original problem via direct substitution or concatenation, serving as useful input information for LLMs. We evaluate the S2L method using both API-based (GPT-4, ChatGPT) and open-source (OpenChat) models over eight symbol-related tasks, ranging from symbol-only abstract reasoning to sentiment analysis in social media. Experimental results show that S2L consistently leads to superior performance. For example, by employing S2L for GPT-4, there can be average significant improvements of +21.9% and +9.5% for subtasks in 1D-ARC and Dyck language, respectively. Codes and data are available at https://github.com/THUNLP-MT/symbol2language.

GMC-IQA: Exploiting Global-correlation and Mean-opinion Consistency for No-reference Image Quality Assessment

Due to the subjective nature of image quality assessment (IQA), assessing which image has better quality among a sequence of images is more reliable than assigning an absolute mean opinion score for an image. Thus, IQA models are evaluated by global correlation consistency (GCC) metrics like PLCC and SROCC, rather than mean opinion consistency (MOC) metrics like MAE and MSE. However, most existing methods adopt MOC metrics to define their loss functions, due to the infeasible computation of GCC metrics during training. In this work, we construct a novel loss function and network to exploit Global-correlation and Mean-opinion Consistency, forming a GMC-IQA framework. Specifically, we propose a novel GCC loss by defining a pairwise preference-based rank estimation to solve the non-differentiable problem of SROCC and introducing a queue mechanism to reserve previous data to approximate the global results of the whole data. Moreover, we propose a mean-opinion network, which integrates diverse opinion features to alleviate the randomness of weight learning and enhance the model robustness. Experiments indicate that our method outperforms SOTA methods on multiple authentic datasets with higher accuracy and generalization. We also adapt the proposed loss to various networks, which brings better performance and more stable training.