LLMs have revolutionized the field of artificial intelligence and have emerged as the de-facto tool for many tasks. The current established technology of LLMs is to process input and generate output at the token level. This is in sharp contrast to humans who operate at multiple levels of abstraction, well beyond single words, to analyze information and to generate creative content. In this paper, we present an attempt at an architecture which operates on an explicit higher-level semantic representation, which we name a concept. Concepts are language- and modality-agnostic and represent a higher level idea or action in a flow. Hence, we build a "Large Concept Model". In this study, as proof of feasibility, we assume that a concept corresponds to a sentence, and use an existing sentence embedding space, SONAR, which supports up to 200 languages in both text and speech modalities. The Large Concept Model is trained to perform autoregressive sentence prediction in an embedding space. We explore multiple approaches, namely MSE regression, variants of diffusion-based generation, and models operating in a quantized SONAR space. These explorations are performed using 1.6B parameter models and training data in the order of 1.3T tokens. We then scale one architecture to a model size of 7B parameters and training data of about 2.7T tokens. We perform an experimental evaluation on several generative tasks, namely summarization and a new task of summary expansion. Finally, we show that our model exhibits impressive zero-shot generalization performance to many languages, outperforming existing LLMs of the same size. The training code of our models is freely available.
The goal of this work was to compute the semantic similarity among publicly available health survey questions in order to facilitate the standardization of survey-based Person-Generated Health Data (PGHD). We compiled various health survey questions authored in both English and Korean from the NIH CDE Repository, PROMIS, Korean public health agencies, and academic publications. Questions were drawn from various health lifelog domains. A randomized question pairing scheme was used to generate a Semantic Text Similarity (STS) dataset consisting of 1758 question pairs. Similarity scores between each question pair were assigned by two human experts. The tagged dataset was then used to build three classifiers featuring: Bag-of-Words, SBERT with BERT-based embeddings, and SBRET with LaBSE embeddings. The algorithms were evaluated using traditional contingency statistics. Among the three algorithms, SBERT-LaBSE demonstrated the highest performance in assessing question similarity across both languages, achieving an Area Under the Receiver Operating Characteristic (ROC) and Precision-Recall Curves of over 0.99. Additionally, it proved effective in identifying cross-lingual semantic similarities.The SBERT-LaBSE algorithm excelled at aligning semantically equivalent sentences across both languages but encountered challenges in capturing subtle nuances and maintaining computational efficiency. Future research should focus on testing with larger multilingual datasets and on calibrating and normalizing scores across the health lifelog domains to improve consistency. This study introduces the SBERT-LaBSE algorithm for calculating semantic similarity across two languages, showing it outperforms BERT-based models and the Bag of Words approach, highlighting its potential to improve semantic interoperability of survey-based PGHD across language barriers.
Sentence embedding models play a key role in various Natural Language Processing tasks, such as in Topic Modeling, Document Clustering and Recommendation Systems. However, these models rely heavily on parallel data, which can be scarce for many low-resource languages, including Luxembourgish. This scarcity results in suboptimal performance of monolingual and cross-lingual sentence embedding models for these languages. To address this issue, we compile a relatively small but high-quality human-generated cross-lingual parallel dataset to train LuxEmbedder, an enhanced sentence embedding model for Luxembourgish with strong cross-lingual capabilities. Additionally, we present evidence suggesting that including low-resource languages in parallel training datasets can be more advantageous for other low-resource languages than relying solely on high-resource language pairs. Furthermore, recognizing the lack of sentence embedding benchmarks for low-resource languages, we create a paraphrase detection benchmark specifically for Luxembourgish, aiming to partially fill this gap and promote further research.
Mining parallel document pairs poses a significant challenge because existing sentence embedding models often have limited context windows, preventing them from effectively capturing document-level information. Another overlooked issue is the lack of concrete evaluation benchmarks comprising high-quality parallel document pairs for assessing document-level mining approaches, particularly for Indic languages. In this study, we introduce Pralekha, a large-scale benchmark for document-level alignment evaluation. Pralekha includes over 2 million documents, with a 1:2 ratio of unaligned to aligned pairs, covering 11 Indic languages and English. Using Pralekha, we evaluate various document-level mining approaches across three dimensions: the embedding models, the granularity levels, and the alignment algorithm. To address the challenge of aligning documents using sentence and chunk-level alignments, we propose a novel scoring method, Document Alignment Coefficient (DAC). DAC demonstrates substantial improvements over baseline pooling approaches, particularly in noisy scenarios, achieving average gains of 20-30% in precision and 15-20% in F1 score. These results highlight DAC's effectiveness in parallel document mining for Indic languages.
This paper presents a novel method to improve the robustness of foundation models to group-based biases. We propose a simple yet effective method, called DoubleCCA, that leverages random sentences and Canonical Correlation Analysis (CCA) to enrich the text embeddings of the foundation model. First, we generate various random sentences that augment the original prompts, which extends the original prompts with random words or character sequences. Second, we use an additional sentence embedding model to generate different text embeddings with respect to these random sentences. We then use CCA double twice to align the representations and reconstruct them back to the original representation space. We demonstrate the effectiveness of our method on a variety of tasks and datasets, showing that it outperforms existing methods in terms of both performance and robustness. Our method is simple to implement and can be easily integrated into existing models, making it a practical solution for improving the robustness of foundation models to group-based biases.
Sentence-level embedding is essential for various tasks that require understanding natural language. Many studies have explored such embeddings for high-resource languages like English. However, low-resource languages like Bengali (a language spoken by almost two hundred and thirty million people) are still under-explored. This work introduces two lightweight sentence transformers for the Bangla language, leveraging a novel cross-lingual knowledge distillation approach. This method distills knowledge from a pre-trained, high-performing English sentence transformer. Proposed models are evaluated across multiple downstream tasks, including paraphrase detection, semantic textual similarity (STS), and Bangla hate speech detection. The new method consistently outperformed existing Bangla sentence transformers. Moreover, the lightweight architecture and shorter inference time make the models highly suitable for deployment in resource-constrained environments, making them valuable for practical NLP applications in low-resource languages.
This research addresses command-line embedding in cybersecurity, a field obstructed by the lack of comprehensive datasets due to privacy and regulation concerns. We propose the first dataset of similar command lines, named CyPHER, for training and unbiased evaluation. The training set is generated using a set of large language models (LLMs) comprising 28,520 similar command-line pairs. Our testing dataset consists of 2,807 similar command-line pairs sourced from authentic command-line data. In addition, we propose a command-line embedding model named CmdCaliper, enabling the computation of semantic similarity with command lines. Performance evaluations demonstrate that the smallest version of CmdCaliper (30 million parameters) suppresses state-of-the-art (SOTA) sentence embedding models with ten times more parameters across various tasks (e.g., malicious command-line detection and similar command-line retrieval). Our study explores the feasibility of data generation using LLMs in the cybersecurity domain. Furthermore, we release our proposed command-line dataset, embedding models' weights and all program codes to the public. This advancement paves the way for more effective command-line embedding for future researchers.
This paper presents the process of building a neural machine translation system with support for English, Romanian, and Aromanian - an endangered Eastern Romance language. The primary contribution of this research is twofold: (1) the creation of the most extensive Aromanian-Romanian parallel corpus to date, consisting of 79,000 sentence pairs, and (2) the development and comparative analysis of several machine translation models optimized for Aromanian. To accomplish this, we introduce a suite of auxiliary tools, including a language-agnostic sentence embedding model for text mining and automated evaluation, complemented by a diacritics conversion system for different writing standards. This research brings contributions to both computational linguistics and language preservation efforts by establishing essential resources for a historically under-resourced language. All datasets, trained models, and associated tools are public: https://huggingface.co/aronlp and https://arotranslate.com
Training-free embedding methods directly leverage pretrained large language models (LLMs) to embed text, bypassing the costly and complex procedure of contrastive learning. Previous training-free embedding methods have mainly focused on optimizing embedding prompts and have overlooked the benefits of utilizing the generative abilities of LLMs. We propose a novel method, GenEOL, which uses LLMs to generate diverse transformations of a sentence that preserve its meaning, and aggregates the resulting embeddings of these transformations to enhance the overall sentence embedding. GenEOL significantly outperforms the existing training-free embedding methods by an average of 2.85 points across several LLMs on the sentence semantic text similarity (STS) benchmark. Our analysis shows that GenEOL stabilizes representation quality across LLM layers and is robust to perturbations of embedding prompts. GenEOL also achieves notable gains on multiple clustering, reranking and pair-classification tasks from the MTEB benchmark.
In virtual assistant (VA) systems it is important to reject or redirect user queries that fall outside the scope of the system. One of the most accurate approaches for out-of-scope (OOS) rejection is to combine it with the task of intent classification on in-scope queries, and to use methods based on the similarity of embeddings produced by transformer-based sentence encoders. Typically, such encoders are fine-tuned for the intent-classification task, using cross-entropy loss. Recent work has shown that while this produces suitable embeddings for the intent-classification task, it also tends to disperse in-scope embeddings over the full sentence embedding space. This causes the in-scope embeddings to potentially overlap with OOS embeddings, thereby making OOS rejection difficult. This is compounded when OOS data is unknown. To mitigate this issue our work proposes to regularize the cross-entropy loss with an in-scope embedding reconstruction loss learned using an auto-encoder. Our method achieves a 1-4% improvement in the area under the precision-recall curve for rejecting out-of-sample (OOS) instances, without compromising intent classification performance.
Scientific document classification is a critical task and often involves many classes. However, collecting human-labeled data for many classes is expensive and usually leads to label-scarce scenarios. Moreover, recent work has shown that sentence embedding model fine-tuning for few-shot classification is efficient, robust, and effective. In this work, we propose FusionSent (Fusion-based Sentence Embedding Fine-tuning), an efficient and prompt-free approach for few-shot classification of scientific documents with many classes. FusionSent uses available training examples and their respective label texts to contrastively fine-tune two different sentence embedding models. Afterward, the parameters of both fine-tuned models are fused to combine the complementary knowledge from the separate fine-tuning steps into a single model. Finally, the resulting sentence embedding model is frozen to embed the training instances, which are then used as input features to train a classification head. Our experiments show that FusionSent significantly outperforms strong baselines by an average of $6.0$ $F_{1}$ points across multiple scientific document classification datasets. In addition, we introduce a new dataset for multi-label classification of scientific documents, which contains 203,961 scientific articles and 130 classes from the arXiv category taxonomy. Code and data are available at https://github.com/sebischair/FusionSent.
Electronic medical records (EMRs) contain the majority of patients' healthcare details. It is an abundant resource for developing an automatic healthcare system. Most of the natural language processing (NLP) studies on EMR processing, such as concept extraction, are adversely affected by the inaccurate segmentation of EMR sections. At the same time, not enough attention has been given to the accurate sectioning of EMRs. The information that may occur in section structures is unvalued. This work focuses on the segmentation of EMRs and proposes a black-box segmentation method using a simple sentence embedding model and neural network, along with a proper training method. To achieve universal adaptivity, we train our model on the dataset with different section headings formats. We compare several advanced deep learning-based NLP methods, and our method achieves the best segmentation accuracies (above 98%) on various test data with a proper training corpus.
The advent of large language models (LLMs) has significantly advanced artificial intelligence (AI) in software engineering (SE), with source code embeddings playing a crucial role in tasks such as source code clone detection and source code clustering. However, existing methods for source code embedding, including those based on LLMs, often rely on costly supervised training or fine-tuning for domain adaptation. This paper proposes a novel approach to embedding source code by combining large language and sentence embedding models. This approach attempts to eliminate the need for task-specific training or fine-tuning and to effectively address the issue of erroneous information commonly found in LLM-generated outputs. To evaluate the performance of our proposed approach, we conducted a series of experiments on three datasets with different programming languages by considering various LLMs and sentence embedding models. The experimental results have demonstrated the effectiveness and superiority of our approach over the state-of-the-art unsupervised approaches, such as SourcererCC, Code2vec, InferCode, TransformCode, and LLM2Vec. Our findings highlight the potential of our approach to advance the field of SE by providing robust and efficient solutions for source code embedding tasks.
This study examines the effectiveness of layer pruning in creating efficient Sentence BERT (SBERT) models. Our goal is to create smaller sentence embedding models that reduce complexity while maintaining strong embedding similarity. We assess BERT models like Muril and MahaBERT-v2 before and after pruning, comparing them with smaller, scratch-trained models like MahaBERT-Small and MahaBERT-Smaller. Through a two-phase SBERT fine-tuning process involving Natural Language Inference (NLI) and Semantic Textual Similarity (STS), we evaluate the impact of layer reduction on embedding quality. Our findings show that pruned models, despite fewer layers, perform competitively with fully layered versions. Moreover, pruned models consistently outperform similarly sized, scratch-trained models, establishing layer pruning as an effective strategy for creating smaller, efficient embedding models. These results highlight layer pruning as a practical approach for reducing computational demand while preserving high-quality embeddings, making SBERT models more accessible for languages with limited technological resources.
Recently, using large language models (LLMs) for data augmentation has led to considerable improvements in unsupervised sentence embedding models. However, existing methods encounter two primary challenges: limited data diversity and high data noise. Current approaches often neglect fine-grained knowledge, such as entities and quantities, leading to insufficient diversity. Additionally, unsupervised data frequently lacks discriminative information, and the generated synthetic samples may introduce noise. In this paper, we propose a pipeline-based data augmentation method via LLMs and introduce the Gaussian-decayed gradient-assisted Contrastive Sentence Embedding (GCSE) model to enhance unsupervised sentence embeddings. To tackle the issue of low data diversity, our pipeline utilizes knowledge graphs (KGs) to extract entities and quantities, enabling LLMs to generate more diverse, knowledge-enriched samples. To address high data noise, the GCSE model uses a Gaussian-decayed function to limit the impact of false hard negative samples, enhancing the model's discriminative capability. Experimental results show that our approach achieves state-of-the-art performance in semantic textual similarity (STS) tasks, using fewer data samples and smaller LLMs, demonstrating its efficiency and robustness across various models.
This paper examines the Code-Switching (CS) phenomenon where two languages intertwine within a single utterance. There exists a noticeable need for research on the CS between English and Korean. We highlight that the current Equivalence Constraint (EC) theory for CS in other languages may only partially capture English-Korean CS complexities due to the intrinsic grammatical differences between the languages. We introduce a novel Koglish dataset tailored for English-Korean CS scenarios to mitigate such challenges. First, we constructed the Koglish-GLUE dataset to demonstrate the importance and need for CS datasets in various tasks. We found the differential outcomes of various foundation multilingual language models when trained on a monolingual versus a CS dataset. Motivated by this, we hypothesized that SimCSE, which has shown strengths in monolingual sentence embedding, would have limitations in CS scenarios. We construct a novel Koglish-NLI (Natural Language Inference) dataset using a CS augmentation-based approach to verify this. From this CS-augmented dataset Koglish-NLI, we propose a unified contrastive learning and augmentation method for code-switched embeddings, ConCSE, highlighting the semantics of CS sentences. Experimental results validate the proposed ConCSE with an average performance enhancement of 1.77\% on the Koglish-STS(Semantic Textual Similarity) tasks.
In an enterprise Virtual Assistant (VA) system, intent classification is the crucial component that determines how a user input is handled based on what the user wants. The VA system is expected to be a cost-efficient SaaS service with low training and inference time while achieving high accuracy even with a small number of training samples. We pretrain a transformer-based sentence embedding model with a contrastive learning objective and leverage the embedding of the model as features when training intent classification models. Our approach achieves the state-of-the-art results for few-shot scenarios and performs better than other commercial solutions on popular intent classification benchmarks. However, generating features via a transformer-based model increases the inference time, especially for longer user inputs, due to the quadratic runtime of the transformer's attention mechanism. On top of model distillation, we introduce a practical multi-task adaptation approach that configures dynamic token pruning without the need for task-specific training for intent classification. We demonstrate that this approach improves the inference speed of popular sentence transformer models without affecting model performance.
Pre-trained transformer models shine in many natural language processing tasks and therefore are expected to bear the representation of the input sentence or text meaning. These sentence-level embeddings are also important in retrieval-augmented generation. But do commonly used plain averaging or prompt templates sufficiently capture and represent the underlying meaning? After providing a comprehensive review of existing sentence embedding extraction and refinement methods, we thoroughly test different combinations and our original extensions of the most promising ones on pretrained models. Namely, given 110 M parameters, BERT's hidden representations from multiple layers, and many tokens, we try diverse ways to extract optimal sentence embeddings. We test various token aggregation and representation post-processing techniques. We also test multiple ways of using a general Wikitext dataset to complement BERT's sentence embeddings. All methods are tested on eight Semantic Textual Similarity (STS), six short text clustering, and twelve classification tasks. We also evaluate our representation-shaping techniques on other static models, including random token representations. Proposed representation extraction methods improve the performance on STS and clustering tasks for all models considered. Very high improvements for static token-based models, especially random embeddings for STS tasks, almost reach the performance of BERT-derived representations. Our work shows that the representation-shaping techniques significantly improve sentence embeddings extracted from BERT-based and simple baseline models.
We propose reCSE, a self supervised contrastive learning sentence representation framework based on feature reshaping. This framework is different from the current advanced models that use discrete data augmentation methods, but instead reshapes the input features of the original sentence, aggregates the global information of each token in the sentence, and alleviates the common problems of representation polarity and GPU memory consumption linear increase in current advanced models. In addition, our reCSE has achieved competitive performance in semantic similarity tasks. And the experiment proves that our proposed feature reshaping method has strong universality, which can be transplanted to other self supervised contrastive learning frameworks and enhance their representation ability, even achieving state-of-the-art performance. Our code is available at https://github.com/heavenhellchen/reCSE.
The ability of generative large language models (LLMs) to perform in-context learning has given rise to a large body of research into how best to prompt models for various natural language processing tasks. In this paper, we focus on machine translation (MT), a task that has been shown to benefit from in-context translation examples. However no systematic studies have been published on how best to select examples, and mixed results have been reported on the usefulness of similarity-based selection over random selection. We provide a study covering multiple LLMs and multiple in-context example retrieval strategies, comparing multilingual sentence embeddings. We cover several language directions, representing different levels of language resourcedness (English into French, German, Swahili and Wolof). Contrarily to previously published results, we find that sentence embedding similarity can improve MT, especially for low-resource language directions, and discuss the balance between selection pool diversity and quality. We also highlight potential problems with the evaluation of LLM-based MT and suggest a more appropriate evaluation protocol, adapting the COMET metric to the evaluation of LLMs. Code and outputs are freely available at https://github.com/ArmelRandy/ICL-MT.