The rapid development of Artificial Intelligence (AI) has revolutionized numerous fields, with large language models (LLMs) and computer vision (CV) systems driving advancements in natural language understanding and visual processing, respectively. The convergence of these technologies has catalyzed the rise of multimodal AI, enabling richer, cross-modal understanding that spans text, vision, audio, and video modalities. Multimodal large language models (MLLMs), in particular, have emerged as a powerful framework, demonstrating impressive capabilities in tasks like image-text generation, visual question answering, and cross-modal retrieval. Despite these advancements, the complexity and scale of MLLMs introduce significant challenges in interpretability and explainability, essential for establishing transparency, trustworthiness, and reliability in high-stakes applications. This paper provides a comprehensive survey on the interpretability and explainability of MLLMs, proposing a novel framework that categorizes existing research across three perspectives: (I) Data, (II) Model, (III) Training \& Inference. We systematically analyze interpretability from token-level to embedding-level representations, assess approaches related to both architecture analysis and design, and explore training and inference strategies that enhance transparency. By comparing various methodologies, we identify their strengths and limitations and propose future research directions to address unresolved challenges in multimodal explainability. This survey offers a foundational resource for advancing interpretability and transparency in MLLMs, guiding researchers and practitioners toward developing more accountable and robust multimodal AI systems.
We introduce OmniFlow, a novel generative model designed for any-to-any generation tasks such as text-to-image, text-to-audio, and audio-to-image synthesis. OmniFlow advances the rectified flow (RF) framework used in text-to-image models to handle the joint distribution of multiple modalities. It outperforms previous any-to-any models on a wide range of tasks, such as text-to-image and text-to-audio synthesis. Our work offers three key contributions: First, we extend RF to a multi-modal setting and introduce a novel guidance mechanism, enabling users to flexibly control the alignment between different modalities in the generated outputs. Second, we propose a novel architecture that extends the text-to-image MMDiT architecture of Stable Diffusion 3 and enables audio and text generation. The extended modules can be efficiently pretrained individually and merged with the vanilla text-to-image MMDiT for fine-tuning. Lastly, we conduct a comprehensive study on the design choices of rectified flow transformers for large-scale audio and text generation, providing valuable insights into optimizing performance across diverse modalities. The Code will be available at https://github.com/jacklishufan/OmniFlows.
The transformer architecture has become an integral part of the field of modern neural networks, playing a crucial role in a variety of tasks, such as text generation, machine translation, image and audio processing, among others. There is also an alternative approach to building intelligent systems, proposed by Jeff Hawkins and inspired by the processes occurring in the neocortex. In our article we want to combine some of these ideas and to propose the use of homeostasis mechanisms, such as RFB-kWTA and "Smart" Inhibition, in the attention mechanism of the transformer and at the output of the transformer block, as well as conducting an experiment involving the introduction of sparse distributed representations of the transformer at various points. RFB-kWTA utilizes statistics of layer activations across time to adjust the entire layer, enhancing the values of rare activations while reducing those of frequent ones. "Smart" Inhibition also uses activation statistics to sample sparsity masks, with rarer activation times are more likely to be activated. Our proposed mechanisms significantly outperform the classical transformer 0.2768 BLEU and a model that only makes use of dropout in the attention mechanism and output of the transformer block 0.3007 BLEU, achieving a score of 0.3062 on the Multi30K dataset.
Prefix circuits are fundamental components in digital adders, widely used in digital systems due to their efficiency in calculating carry signals. Synthesizing prefix circuits with minimized area and delay is crucial for enhancing the performance of modern computing systems. Recently, large language models (LLMs) have demonstrated a surprising ability to perform text generation tasks. We propose PrefixLLM, that leverages LLMs for prefix circuit synthesis. PrefixLLM transforms the prefix circuit synthesis task into a structured text generation problem, termed the Structured Prefix Circuit Representation (SPCR), and introduces an iterative framework to automatically and accurately generate valid SPCRs. We further present a design space exploration (DSE) framework that uses LLMs to iteratively search for area and delay optimized prefix circuits. Compared to state-of-the-art, PrefixLLM can reduce the area by 3.70% under the same delay constraint. This work highlights the use of LLMs in the synthesis of arithmetic circuits, which can be transformed into the structured text generation.
Programmable Logic Controllers (PLCs) are microcomputers essential for automating factory operations. Structured Text (ST), a high-level language adhering to the IEC 61131-3 standard, is pivotal for PLCs due to its ability to express logic succinctly and to seamlessly integrate with other languages within the same standard. However, vendors develop their own customized versions of ST, and the lack of comprehensive and standardized documentation for the full semantics of ST has contributed to inconsistencies in how the language is implemented. Consequently, the steep learning curve associated with ST, combined with ever-evolving industrial requirements, presents significant challenges for developers. In response to these issues, we present AutoPLC, an LLM-based approach designed to automate the generation of vendor-specific ST code. To facilitate effective code generation, we first built a comprehensive knowledge base, including Rq2ST Case Library (requirements and corresponding implementations) and Instruction libraries. Then we developed a retrieval module to incorporate the domain-specific knowledge by identifying pertinent cases and instructions, guiding the LLM to generate code that meets the requirements. In order to verify and improve the quality of the generated code, we designed an adaptable code checker. If errors are detected, we initiate an iterative self-improvement process to instruct the LLM to revise the generated code. We evaluate AutoPLC's performance against seven state-of-the-art baselines using three benchmarks, one for open-source basic ST and two for commercial Structured Control Language (SCL) from Siemens. The results show that our approach consistently achieves superior performance across all benchmarks. Ablation study emphasizes the significance of our modules. Further manual analysis confirm the practical utility of the ST code generated by AutoPLC.
Language models based on deep neural networks are vulnerable to textual adversarial attacks. While rich-resource languages like English are receiving focused attention, Tibetan, a cross-border language, is gradually being studied due to its abundant ancient literature and critical language strategy. Currently, there are several Tibetan adversarial text generation methods, but they do not fully consider the textual features of Tibetan script and overestimate the quality of generated adversarial texts. To address this issue, we propose a novel Tibetan adversarial text generation method called TSCheater, which considers the characteristic of Tibetan encoding and the feature that visually similar syllables have similar semantics. This method can also be transferred to other abugidas, such as Devanagari script. We utilize a self-constructed Tibetan syllable visual similarity database called TSVSDB to generate substitution candidates and adopt a greedy algorithm-based scoring mechanism to determine substitution order. After that, we conduct the method on eight victim language models. Experimentally, TSCheater outperforms existing methods in attack effectiveness, perturbation magnitude, semantic similarity, visual similarity, and human acceptance. Finally, we construct the first Tibetan adversarial robustness evaluation benchmark called AdvTS, which is generated by existing methods and proofread by humans.
After the introduction of Large Language Models (LLMs), there have been substantial improvements in the performance of Natural Language Generation (NLG) tasks, including Text Summarization and Machine Translation. However, LLMs still produce outputs containing hallucinations, that is, content not grounded in factual information. Therefore, developing methods to assess the factuality of LLMs has become urgent. Indeed, resources for factuality evaluation have recently emerged. Although challenging, these resources face one or more of the following limitations: (i) they are tailored to a specific task or domain; (ii) they are limited in size, thereby preventing the training of new factuality evaluators; (iii) they are designed for simpler verification tasks, such as claim verification. To address these issues, we introduce LLM-Oasis, to the best of our knowledge the largest resource for training end-to-end factuality evaluators. LLM-Oasis is constructed by extracting claims from Wikipedia, falsifying a subset of these claims, and generating pairs of factual and unfactual texts. We then rely on human annotators to both validate the quality of our dataset and to create a gold standard test set for benchmarking factuality evaluation systems. Our experiments demonstrate that LLM-Oasis presents a significant challenge for state-of-the-art LLMs, with GPT-4o achieving up to 60% accuracy in our proposed end-to-end factuality evaluation task, highlighting its potential to drive future research in the field.
Large Language Models (LLMs) have shown exceptional performance across various Data-to-Text Generation (DTG) tasks. However, generating factually consistent text in DTG remains challenging for LLMs. Despite this, in-depth evaluations of LLM factual consistency for DTG remain missing in the current literature. This paper addresses this gap by providing an extensive evaluation of factual consistency in LLMs for DTG. Our evaluation covers five widely used DTG datasets (E2E, ViGGo, WikiTableText, DART, and WebNLG) and five prominent LLM families (T5, BART, OPT, BLOOM, and Llama 2). To ensure a thorough evaluation of factual consistency, we use four state-of-the-art automatic metrics and include essential human assessments. Our extensive evaluations reveals three key findings regarding factual consistency in LLMs for DTG. First, Llama 2 often excels in generating factually consistent text, although smaller models like T5 and BART can achieve strong factual consistency on larger, lexically less-diverse datasets. Second, the average rate of change (AROC) indicates that increasing model size (number of model trainable parameters) generally enhances factual consistency of LLMs in DTG. Third, we observe that source-reference divergence (i.e., when the reference text diverges semantically from the source) typically reduces the factual consistency of LLMs in DTG.
The SQL-to-text generation task traditionally uses template base, Seq2Seq, tree-to-sequence, and graph-to-sequence models. Recent models take advantage of pre-trained generative language models for this task in the Seq2Seq framework. However, treating SQL as a sequence of inputs to the pre-trained models is not optimal. In this work, we put forward a new SQL intermediate representation called EzSQL to align SQL with the natural language text sequence. EzSQL simplifies the SQL queries and brings them closer to natural language text by modifying operators and keywords, which can usually be described in natural language. EzSQL also removes the need for set operators. Our proposed SQL-to-text generation model uses EzSQL as the input to a pre-trained generative language model for generating the text descriptions. We demonstrate that our model is an effective state-of-the-art method to generate text narrations from SQL queries on the WikiSQL and Spider datasets. We also show that by generating pretraining data using our SQL-to-text generation model, we can enhance the performance of Text-to-SQL parsers.
Large Language Models (LLMs) have demonstrated remarkable multilingual capabilities, yet challenges persist in adapting these models for low-resource languages. In this study, we investigate the effects of Low-Rank Adaptation (LoRA) Parameter-Efficient Fine-Tuning (PEFT) on multilingual Gemma models for Marathi, a language with limited resources. Using a translated Alpaca dataset with 52,000 instruction-response pairs, our findings reveal that while evaluation metrics often show a performance decline post-fine-tuning, manual assessments frequently suggest that the fine-tuned models outperform their original counterparts. The observations indicate improvements in target language generation capabilities but a reduction in reasoning abilities following language adaptation. These results underscore the need for improved evaluation methodologies and the creation of high-quality native datasets to accurately assess language-specific model performance in low-resource settings.
Sign language is a visual language that encompasses all linguistic features of natural languages and serves as the primary communication method for the deaf and hard-of-hearing communities. While many studies have successfully adapted pretrained language models (LMs) for sign language translation (sign-to-text), drawing inspiration from its linguistic characteristics, the reverse task of sign language generation (SLG, text-to-sign) remains largely unexplored. Most existing approaches treat SLG as a visual content generation task, employing techniques such as diffusion models to produce sign videos, 2D keypoints, or 3D avatars based on text inputs, overlooking the linguistic properties of sign languages. In this work, we introduce a multilingual sign language model, Signs as Tokens (SOKE), which can generate 3D sign avatars autoregressively from text inputs using a pretrained LM. To align sign language with the LM, we develop a decoupled tokenizer that discretizes continuous signs into token sequences representing various body parts. These sign tokens are integrated into the raw text vocabulary of the LM, allowing for supervised fine-tuning on sign language datasets. To facilitate multilingual SLG research, we further curate a large-scale Chinese sign language dataset, CSL-Daily, with high-quality 3D pose annotations. Extensive qualitative and quantitative evaluations demonstrate the effectiveness of SOKE. The project page is available at https://2000zrl.github.io/soke/.
Multimodal Large Language Models (MLLMs) have made significant strides in visual understanding and generation tasks. However, generating interleaved image-text content remains a challenge, which requires integrated multimodal understanding and generation abilities. While the progress in unified models offers new solutions, existing benchmarks are insufficient for evaluating these methods due to limitations in data size and diversity. To bridge this gap, we introduce OpenING, a comprehensive benchmark comprising 5,400 high-quality human-annotated instances across 56 real-world tasks. OpenING covers diverse daily scenarios such as travel guide, design, and brainstorming, offering a robust platform for challenging interleaved generation methods. In addition, we present IntJudge, a judge model for evaluating open-ended multimodal generation methods. Trained with a novel data pipeline, our IntJudge achieves an agreement rate of 82.42% with human judgments, outperforming GPT-based evaluators by 11.34%. Extensive experiments on OpenING reveal that current interleaved generation methods still have substantial room for improvement. Key findings on interleaved image-text generation are further presented to guide the development of next-generation models.
The quality of meeting summaries generated by natural language generation (NLG) systems is hard to measure automatically. Established metrics such as ROUGE and BERTScore have a relatively low correlation with human judgments and fail to capture nuanced errors. Recent studies suggest using large language models (LLMs), which have the benefit of better context understanding and adaption of error definitions without training on a large number of human preference judgments. However, current LLM-based evaluators risk masking errors and can only serve as a weak proxy, leaving human evaluation the gold standard despite being costly and hard to compare across studies. In this work, we present MESA, an LLM-based framework employing a three-step assessment of individual error types, multi-agent discussion for decision refinement, and feedback-based self-training to refine error definition understanding and alignment with human judgment. We show that MESA's components enable thorough error detection, consistent rating, and adaptability to custom error guidelines. Using GPT-4o as its backbone, MESA achieves mid to high Point-Biserial correlation with human judgment in error detection and mid Spearman and Kendall correlation in reflecting error impact on summary quality, on average 0.25 higher than previous methods. The framework's flexibility in adapting to custom error guidelines makes it suitable for various tasks with limited human-labeled data.
Knowledge distillation (KD) has become a widely adopted approach for compressing large language models (LLMs) to reduce computational costs and memory footprints. However, the availability of complex teacher models is a prerequisite for running most KD pipelines. Thus, the traditional KD procedure can be unachievable or budget-unfriendly, particularly when relying on commercial LLMs like GPT4. In this regard, Self-distillation (SelfD) emerges as an advisable alternative, enabling student models to learn without teachers' guidance. Nonetheless, existing SelfD approaches for LMs often involve architectural modifications, assuming the models are open-source, which may not always be practical. In this work, we introduce a model-agnostic and task-agnostic method named dynamic SelfD from the previous minibatch (DynSDPB), which realizes current iterations' distillation from the last ones' generated logits. Additionally, to address prediction inaccuracies during the early iterations, we dynamically adjust the distillation influence and temperature values to enhance the adaptability of fine-tuning. Furthermore, DynSDPB is a novel fine-tuning policy that facilitates the seamless integration of existing self-correction and self-training techniques for small language models (SLMs) because they all require updating SLMs' parameters. We demonstrate the superior performance of DynSDPB on both encoder-only LMs (e.g., BERT model families) and decoder-only LMs (e.g., LLaMA model families), validating its effectiveness across natural language understanding (NLU) and natural language generation (NLG) benchmarks.
Non-autoregressive language models generate all tokens simultaneously, offering potential speed advantages over traditional autoregressive models, but they face challenges in modeling the complex dependencies inherent in text data. In this work, we investigate a conditional flow matching approach for text generation. We represent tokens as one-hot vectors in a \(V\)-dimensional simplex and utilize geodesics under the Kullback-Leibler (KL) divergence, which correspond to linear interpolation in logit space. We provide a theoretical justification that maximizing the conditional likelihood \(P_{\theta}(x_1 \mid x_t, t)\) yields the exact flow matching velocity under logit interpolation. To address the suboptimal performance of basic inference, we propose a novel empirical sampling scheme that iteratively samples from the conditional distribution and introduces additional noise, significantly improving results despite lacking full theoretical underpinnings. Furthermore, we propose a hybrid inference method that combines the basic approach with the sampling scheme. This method demonstrates superior performance on both conditional and unconditional text generation experiments compared to previous SOTA method for discrete flow matching.
Recent advancements in Virtual Try-On (VTO) have demonstrated exceptional efficacy in generating realistic images and preserving garment details, largely attributed to the robust generative capabilities of text-to-image (T2I) diffusion backbones. However, the T2I models that underpin these methods have become outdated, thereby limiting the potential for further improvement in VTO. Additionally, current methods face notable challenges in accurately rendering text on garments without distortion and preserving fine-grained details, such as textures and material fidelity. The emergence of Diffusion Transformer (DiT) based T2I models has showcased impressive performance and offers a promising opportunity for advancing VTO. Directly applying existing VTO techniques to transformer-based T2I models is ineffective due to substantial architectural differences, which hinder their ability to fully leverage the models' advanced capabilities for improved text generation. To address these challenges and unlock the full potential of DiT-based T2I models for VTO, we propose TED-VITON, a novel framework that integrates a Garment Semantic (GS) Adapter for enhancing garment-specific features, a Text Preservation Loss to ensure accurate and distortion-free text rendering, and a constraint mechanism to generate prompts by optimizing Large Language Model (LLM). These innovations enable state-of-the-art (SOTA) performance in visual quality and text fidelity, establishing a new benchmark for VTO task. Project page: \url{https://zhenchenwan.github.io/TED-VITON/}
The incivility in social media discourse complicates the deployment of automated text generation models for politically sensitive content. Fine-tuning and prompting strategies are critical, but underexplored, solutions to mitigate toxicity in such contexts. This study investigates the fine-tuning and prompting effects on GPT-3.5 Turbo using subsets of the CLAPTON dataset of political discussion posts, comprising Twitter and Reddit data labeled for their justification, reciprocity and incivility. Fine-tuned models on Reddit data scored highest on discussion quality, while combined noisy data led to persistent toxicity. Prompting strategies reduced specific toxic traits, such as personal attacks, but had limited broader impact. The findings emphasize that high-quality data and well-crafted prompts are essential to reduce incivility and improve rhetorical quality in automated political discourse generation.
With the recent development of natural language generation models - termed as large language models (LLMs) - a potential use case has opened up to improve the way that humans interact with robot assistants. These LLMs should be able to leverage their large breadth of understanding to interpret natural language commands into effective, task appropriate and safe robot task executions. However, in reality, these models suffer from hallucinations, which may cause safety issues or deviations from the task. In other domains, these issues have been improved through the use of collaborative AI systems where multiple LLM agents can work together to collectively plan, code and self-check outputs. In this research, multiple collaborative AI systems were tested against a single independent AI agent to determine whether the success in other domains would translate into improved human-robot interaction performance. The results show that there is no defined trend between the number of agents and the success of the model. However, it is clear that some collaborative AI agent architectures can exhibit a greatly improved capacity to produce error-free code and to solve abstract problems.
As the text-to-image (T2I) domain progresses, generating text that seamlessly integrates with visual content has garnered significant attention. However, even with accurate text generation, the inability to control font and color can greatly limit certain applications, and this issue remains insufficiently addressed. This paper introduces AnyText2, a novel method that enables precise control over multilingual text attributes in natural scene image generation and editing. Our approach consists of two main components. First, we propose a WriteNet+AttnX architecture that injects text rendering capabilities into a pre-trained T2I model. Compared to its predecessor, AnyText, our new approach not only enhances image realism but also achieves a 19.8% increase in inference speed. Second, we explore techniques for extracting fonts and colors from scene images and develop a Text Embedding Module that encodes these text attributes separately as conditions. As an extension of AnyText, this method allows for customization of attributes for each line of text, leading to improvements of 3.3% and 9.3% in text accuracy for Chinese and English, respectively. Through comprehensive experiments, we demonstrate the state-of-the-art performance of our method. The code and model will be made open-source in https://github.com/tyxsspa/AnyText2.
Most current molecular language models transfer the masked language model or image-text generation model from natural language processing to molecular field. However, molecules are not solely characterized by atom/bond symbols; they encapsulate important physical/chemical properties. Moreover, normal language models bring grammar rules that are irrelevant for understanding molecules. In this study, we propose a novel physicochemical knowledge-guided molecular meta language framework MolMetaLM. We design a molecule-specialized meta language paradigm, formatted as multiple <S,P,O> (subject, predicate, object) knowledge triples sharing the same S (i.e., molecule) to enhance learning the semantic relationships between physicochemical knowledge and molecules. By introducing different molecular knowledge and noises, the meta language paradigm generates tens of thousands of pretraining tasks. By recovering the token/sequence/order-level noises, MolMetaLM exhibits proficiency in large-scale benchmark evaluations involving property prediction, molecule generation, conformation inference, and molecular optimization. Through MolMetaLM, we offer a new insight for designing language models.