We present Adjacent Possible Exploration (APE), a selective fine-tuning method for adapting large language models that systematically explores parameter modifications while maintaining model stability. Inspired by evolutionary optimization principles, APE evaluates multiple candidate parameter updates through fine-tuning on small data subsets and accepts only those exceeding a performance threshold. Unlike standard fine-tuning that follows single gradient directions, APE implements a filtered selection process that prevents destabilizing parameter changes while enabling systematic improvement. Our method achieves 33.9\% BLEU improvement and 36.2\% perplexity reduction on news summarization tasks while using minimal computational resources. The approach provides a practical framework for controlled model adaptation that balances performance gains with representational stability.
Verifying the credibility of Cyber Threat Intelligence (CTI) is essential for reliable cybersecurity defense. However, traditional approaches typically treat this task as a static classification problem, relying on handcrafted features or isolated deep learning models. These methods often lack the robustness needed to handle incomplete, heterogeneous, or noisy intelligence, and they provide limited transparency in decision-making-factors that reduce their effectiveness in real-world threat environments. To address these limitations, we propose LRCTI, a Large Language Model (LLM)-based framework designed for multi-step CTI credibility verification. The framework first employs a text summarization module to distill complex intelligence reports into concise and actionable threat claims. It then uses an adaptive multi-step evidence retrieval mechanism that iteratively identifies and refines supporting information from a CTI-specific corpus, guided by LLM feedback. Finally, a prompt-based Natural Language Inference (NLI) module is applied to evaluate the credibility of each claim while generating interpretable justifications for the classification outcome. Experiments conducted on two benchmark datasets, CTI-200 and PolitiFact show that LRCTI improves F1-Macro and F1-Micro scores by over 5%, reaching 90.9% and 93.6%, respectively, compared to state-of-the-art baselines. These results demonstrate that LRCTI effectively addresses the core limitations of prior methods, offering a scalable, accurate, and explainable solution for automated CTI credibility verification
To advance biomedical vison-language model capabilities through scaling up, fine-tuning, and instruction tuning, develop vision-language models with improved performance in handling long text, explore strategies to efficiently adopt vision language models for diverse multi-modal biomedical tasks, and examine the zero-shot learning performance. We developed two biomedical vision language models, BiomedGPT-Large and BiomedGPT-XLarge, based on an encoder-decoder-based transformer architecture. We fine-tuned the two models on 23 benchmark datasets from 6 multi-modal biomedical tasks including one image-only task (image classification), three language-only tasks (text understanding, text summarization and question answering), and two vision-language tasks (visual question answering and image captioning). We compared the developed scaled models with our previous BiomedGPT-Base model and existing prestigious models reported in the literature. We instruction-tuned the two models using a large-scale multi-modal biomedical instruction-tuning dataset and assessed the zero-shot learning performance and alignment accuracy.
Large language models (LLMs) excel at capturing global token dependencies via self-attention but face prohibitive compute and memory costs on lengthy inputs. While sub-quadratic methods (e.g., linear attention) can reduce these costs, they often degrade accuracy due to overemphasizing recent tokens. In this work, we first propose dual-state linear attention (DSLA), a novel design that maintains two specialized hidden states-one for preserving historical context and one for tracking recency-thereby mitigating the short-range bias typical of linear-attention architectures. To further balance efficiency and accuracy under dynamic workload conditions, we introduce DSLA-Serve, an online adaptive distillation framework that progressively replaces Transformer layers with DSLA layers at inference time, guided by a sensitivity-based layer ordering. DSLA-Serve uses a chained fine-tuning strategy to ensure that each newly converted DSLA layer remains consistent with previously replaced layers, preserving the overall quality. Extensive evaluations on commonsense reasoning, long-context QA, and text summarization demonstrate that DSLA-Serve yields 2.3x faster inference than Llama2-7B and 3.0x faster than the hybrid Zamba-7B, while retaining comparable performance across downstream tasks. Our ablation studies show that DSLA's dual states capture both global and local dependencies, addressing the historical-token underrepresentation seen in prior linear attentions. Codes are available at https://github.com/utnslab/DSLA-Serve.
Large language models (LLMs) have become the cornerstone of modern AI. However, the existing paradigm of next-token prediction fundamentally limits their ability to form coherent, high-level concepts, making it a critical barrier to human-like understanding and reasoning. Take the phrase "ribonucleic acid" as an example: an LLM will first decompose it into tokens, i.e., artificial text fragments ("rib", "on", ...), then learn each token sequentially, rather than grasping the phrase as a unified, coherent semantic entity. This fragmented representation hinders deeper conceptual understanding and, ultimately, the development of truly intelligent systems. In response, we introduce Concept-Aware Fine-Tuning (CAFT), a novel multi-token training method that redefines how LLMs are fine-tuned. By enabling the learning of sequences that span multiple tokens, this method fosters stronger concept-aware learning. Our experiments demonstrate significant improvements compared to conventional next-token finetuning methods across diverse tasks, including traditional applications like text summarization and domain-specific ones like de novo protein design. Multi-token prediction was previously only possible in the prohibitively expensive pretraining phase; CAFT, to our knowledge, is the first to bring the multi-token setting to the post-training phase, thus effectively democratizing its benefits for the broader community of practitioners and researchers. Finally, the unexpected effectiveness of our proposed method suggests wider implications for the machine learning research community. All code and data are available at https://github.com/michaelchen-lab/caft-llm
We present a new adaptation method MaCP, Minimal yet Mighty adaptive Cosine Projection, that achieves exceptional performance while requiring minimal parameters and memory for fine-tuning large foundation models. Its general idea is to exploit the superior energy compaction and decorrelation properties of cosine projection to improve both model efficiency and accuracy. Specifically, it projects the weight change from the low-rank adaptation into the discrete cosine space. Then, the weight change is partitioned over different levels of the discrete cosine spectrum, and each partition's most critical frequency components are selected. Extensive experiments demonstrate the effectiveness of MaCP across a wide range of single-modality tasks, including natural language understanding, natural language generation, text summarization, as well as multi-modality tasks such as image classification and video understanding. MaCP consistently delivers superior accuracy, significantly reduced computational complexity, and lower memory requirements compared to existing alternatives.
This study presents FiLLM, a Filipino-optimized large language model, designed to enhance natural language processing (NLP) capabilities in the Filipino language. Built upon the SeaLLM-7B 2.5 model, FiLLM leverages Low-Rank Adaptation (LoRA) fine-tuning to optimize memory efficiency while maintaining task-specific performance. The model was trained and evaluated on diverse Filipino datasets to address key NLP tasks, including Named Entity Recognition (NER), Part-of-Speech (POS) tagging, Dependency Parsing, and Text Summarization. Performance comparisons with the CalamanCy model were conducted using F1 Score, Precision, Recall, Compression Rate, and Keyword Overlap metrics. Results indicate that Calamancy outperforms FILLM in several aspects, demonstrating its effectiveness in processing Filipino text with improved linguistic comprehension and adaptability. This research contributes to the advancement of Filipino NLP applications by providing an optimized, efficient, and scalable language model tailored for local linguistic needs.
The continued rise of neural networks and large language models in the more recent past has altered the natural language processing landscape, enabling new approaches towards typical language tasks and achieving mainstream success. Despite the huge success of large language models, many disadvantages still remain and through this work we assess the state of the art in five application scenarios with a particular focus on the future perspectives and sensible application scenarios of traditional and older approaches and techniques. In this paper we survey recent publications in the application scenarios classification, information and relation extraction, text simplification as well as text summarization. After defining our terminology, i.e., which features are characteristic for traditional techniques in our interpretation for the five scenarios, we survey if such traditional approaches are still being used, and if so, in what way they are used. It turns out that all five application scenarios still exhibit traditional models in one way or another, as part of a processing pipeline, as a comparison/baseline to the core model of the respective paper, or as the main model(s) of the paper. For the complete statistics, see https://zenodo.org/records/13683801
$\texttt{BIGBIRD-PEGASUS}$ model achieves $\textit{state-of-the-art}$ on abstractive text summarization for long documents. However it's capacity still limited to maximum of $4,096$ tokens, thus caused performance degradation on summarization for very long documents. Common method to deal with the issue is to truncate the documents. In this reasearch, we'll use different approach. We'll use the pretrained $\texttt{BIGBIRD-PEGASUS}$ model by fine tuned the model on other domain dataset. First, we filter out all documents which length less than $20,000$ tokens to focus on very long documents. To prevent domain shifting problem and overfitting on transfer learning due to small dataset, we augment the dataset by splitting document-summary training pair into parts, to fit the document into $4,096$ tokens. Source code available on $\href{https://github.com/lhfazry/SPIN-summ}{https://github.com/lhfazry/SPIN-summ}$.
Vision-Language Models (VLMs) often struggle to balance visual and textual information when summarizing complex multimodal inputs, such as entire TV show episodes. In this paper, we propose a zero-shot video-to-text summarization approach that builds its own screenplay representation of an episode, effectively integrating key video moments, dialogue, and character information into a unified document. Unlike previous approaches, we simultaneously generate screenplays and name the characters in zero-shot, using only the audio, video, and transcripts as input. Additionally, we highlight that existing summarization metrics can fail to assess the multimodal content in summaries. To address this, we introduce MFactSum, a multimodal metric that evaluates summaries with respect to both vision and text modalities. Using MFactSum, we evaluate our screenplay summaries on the SummScreen3D dataset, demonstrating superiority against state-of-the-art VLMs such as Gemini 1.5 by generating summaries containing 20% more relevant visual information while requiring 75% less of the video as input.
The Cancer Genome Atlas (TCGA) has enabled novel discoveries and served as a large-scale reference through its harmonized genomics, clinical, and image data. Prior studies have trained bespoke cancer survival prediction models from unimodal or multimodal TCGA data. A modern paradigm in biomedical deep learning is the development of foundation models (FMs) to derive meaningful feature embeddings, agnostic to a specific modeling task. Biomedical text especially has seen growing development of FMs. While TCGA contains free-text data as pathology reports, these have been historically underutilized. Here, we investigate the feasibility of training classical, multimodal survival models over zero-shot embeddings extracted by FMs. We show the ease and additive effect of multimodal fusion, outperforming unimodal models. We demonstrate the benefit of including pathology report text and rigorously evaluate the effect of model-based text summarization and hallucination. Overall, we modernize survival modeling by leveraging FMs and information extraction from pathology reports.
Unstructured clinical data can serve as a unique and rich source of information that can meaningfully inform clinical practice. Extracting the most pertinent context from such data is critical for exploiting its true potential toward optimal and timely decision-making in patient care. While prior research has explored various methods for clinical text summarization, most prior studies either process all input tokens uniformly or rely on heuristic-based filters, which can overlook nuanced clinical cues and fail to prioritize information critical for decision-making. In this study, we propose Contextual, a novel framework that integrates a Context-Preserving Token Filtering method with a Domain-Specific Knowledge Graph (KG) for contextual augmentation. By preserving context-specific important tokens and enriching them with structured knowledge, ConTextual improves both linguistic coherence and clinical fidelity. Our extensive empirical evaluations on two public benchmark datasets demonstrate that ConTextual consistently outperforms other baselines. Our proposed approach highlights the complementary role of token-level filtering and structured retrieval in enhancing both linguistic and clinical integrity, as well as offering a scalable solution for improving precision in clinical text generation.
Recommender systems have rapidly evolved and become integral to many online services. However, existing systems sometimes produce unstable and unsatisfactory recommendations that fail to align with users' fundamental and long-term preferences. This is because they primarily focus on extracting shallow and short-term interests from user behavior data, which is inherently dynamic and challenging to model. Unlike these transient interests, user values are more stable and play a crucial role in shaping user behaviors, such as purchasing items and consuming content. Incorporating user values into recommender systems can help stabilize recommendation performance and ensure results better reflect users' latent preferences. However, acquiring user values is typically difficult and costly. To address this challenge, we leverage the strong language understanding, zero-shot inference, and generalization capabilities of Large Language Models (LLMs) to extract user values from users' historical interactions. Unfortunately, direct extraction using LLMs presents several challenges such as length constraints and hallucination. To overcome these issues, we propose ZOOM, a zero-shot multi-LLM collaborative framework for effective and accurate user value extraction. In ZOOM, we apply text summarization techniques to condense item content while preserving essential meaning. To mitigate hallucinations, ZOOM introduces two specialized agent roles: evaluators and supervisors, to collaboratively generate accurate user values. Extensive experiments on two widely used recommendation datasets with two state-of-the-art recommendation models demonstrate the effectiveness and generalization of our framework in automatic user value mining and recommendation performance improvement.
Speculative decoding accelerates large language model inference by using smaller draft models to generate candidate tokens for parallel verification. However, current approaches are limited by sequential stage dependencies that prevent full hardware utilization. We present PipeSpec, a framework that generalizes speculative decoding to $k$ models arranged in a hierarchical pipeline, enabling asynchronous execution with lightweight coordination for prediction verification and rollback. Our analytical model characterizes token generation rates across pipeline stages and proves guaranteed throughput improvements over traditional decoding for any non-zero acceptance rate. We further derive closed-form expressions for steady-state verification probabilities that explain the empirical benefits of pipeline depth. Experimental results show that PipeSpec achieves up to 2.54$\times$ speedup while outperforming state-of-the-art methods. We validate PipeSpec across text summarization and code generation tasks using LLaMA 2 and 3 models, demonstrating that pipeline efficiency increases with model depth, providing a scalable approach to accelerating LLM inference on multi-device systems.
With over 700 stars on GitHub and being part of the official ONNX repository, the ONNX Optimizer is the default tool for applying graph-based optimizations to ONNX models. Despite its widespread use, its ability to maintain model accuracy during optimization has not been thoroughly investigated. In this work, we present OODTE, a utility designed to automatically and comprehensively evaluate the correctness of the ONNX Optimizer. OODTE adopts a straightforward yet powerful differential testing and evaluation methodology, which can be readily adapted for use with other compiler optimizers. Specifically, OODTE takes a collection of ONNX models, applies optimizations, and executes both the original and optimized versions across a user-defined input set, automatically capturing any issues encountered during optimization. When discrepancies in accuracy arise, OODTE iteratively isolates the responsible optimization pass by repeating the process at a finer granularity. We applied OODTE to 130 well-known models from the official ONNX Model Hub, spanning diverse tasks including classification, object detection, semantic segmentation, text summarization, question answering, and sentiment analysis. Our evaluation revealed that 9.2% of the model instances either caused the optimizer to crash or led to the generation of invalid models using default optimization strategies. Additionally, 30% of classification models and 16.6% of object detection and segmentation models exhibited differing outputs across original and optimized versions, whereas models focused on text-related tasks were generally robust to optimization. OODTE uncovered 15 issues-14 previously unknown-affecting 9 of 47 optimization passes and the optimizer overall. All issues were reported to the ONNX Optimizer team. OODTE offers a simple but effective framework for validating AI model optimizers, applicable beyond the ONNX ecosystem.
Evaluating text summarization quality remains a critical challenge in Natural Language Processing. Current approaches face a trade-off between performance and interpretability. We present SEval-Ex, a framework that bridges this gap by decomposing summarization evaluation into atomic statements, enabling both high performance and explainability. SEval-Ex employs a two-stage pipeline: first extracting atomic statements from text source and summary using LLM, then a matching between generated statements. Unlike existing approaches that provide only summary-level scores, our method generates detailed evidence for its decisions through statement-level alignments. Experiments on the SummEval benchmark demonstrate that SEval-Ex achieves state-of-the-art performance with 0.580 correlation on consistency with human consistency judgments, surpassing GPT-4 based evaluators (0.521) while maintaining interpretability. Finally, our framework shows robustness against hallucination.
Extracting concise information from scientific documents aids learners, researchers, and practitioners. Automatic Text Summarization (ATS), a key Natural Language Processing (NLP) application, automates this process. While ATS methods exist for many languages, Kurdish remains underdeveloped due to limited resources. This study develops a dataset and language model based on 231 scientific papers in Sorani Kurdish, collected from four academic departments in two universities in the Kurdistan Region of Iraq (KRI), averaging 26 pages per document. Using Sentence Weighting and Term Frequency-Inverse Document Frequency (TF-IDF) algorithms, two experiments were conducted, differing in whether the conclusions were included. The average word count was 5,492.3 in the first experiment and 5,266.96 in the second. Results were evaluated manually and automatically using ROUGE-1, ROUGE-2, and ROUGE-L metrics, with the best accuracy reaching 19.58%. Six experts conducted manual evaluations using three criteria, with results varying by document. This research provides valuable resources for Kurdish NLP researchers to advance ATS and related fields.
LLM inference is essential for applications like text summarization, translation, and data analysis, but the high cost of GPU instances from Cloud Service Providers (CSPs) like AWS is a major burden. This paper proposes InferSave, a cost-efficient VM selection framework for cloud based LLM inference. InferSave optimizes KV cache offloading based on Service Level Objectives (SLOs) and workload charac teristics, estimating GPU memory needs, and recommending cost-effective VM instances. Additionally, the Compute Time Calibration Function (CTCF) improves instance selection accuracy by adjusting for discrepancies between theoretical and actual GPU performance. Experiments on AWS GPU instances show that selecting lower-cost instances without KV cache offloading improves cost efficiency by up to 73.7% for online workloads, while KV cache offloading saves up to 20.19% for offline workloads.
Reasoning-enabled large language models (LLMs) have recently demonstrated impressive performance in complex logical and mathematical tasks, yet their effectiveness in evaluating natural language generation remains unexplored. This study systematically compares reasoning-based LLMs (DeepSeek-R1 and OpenAI o3) with their non-reasoning counterparts across machine translation (MT) and text summarization (TS) evaluation tasks. We evaluate eight models across three architectural categories, including state-of-the-art reasoning models, their distilled variants (ranging from 8B to 70B parameters), and equivalent conventional, non-reasoning LLMs. Our experiments on WMT23 and SummEval benchmarks reveal that the benefits of reasoning capabilities are highly model and task-dependent: while OpenAI o3-mini models show consistent performance improvements with increased reasoning intensity, DeepSeek-R1 underperforms compared to its non-reasoning variant, with exception to certain aspects of TS evaluation. Correlation analysis demonstrates that increased reasoning token usage positively correlates with evaluation quality in o3-mini models. Furthermore, our results show that distillation of reasoning capabilities maintains reasonable performance in medium-sized models (32B) but degrades substantially in smaller variants (8B). This work provides the first comprehensive assessment of reasoning LLMs for NLG evaluation and offers insights into their practical use.
Speech summarization has become an essential tool for efficiently managing and accessing the growing volume of spoken and audiovisual content. However, despite its increasing importance, speech summarization remains loosely defined. The field intersects with several research areas, including speech recognition, text summarization, and specific applications like meeting summarization. This survey not only examines existing datasets and evaluation protocols, which are crucial for assessing the quality of summarization approaches, but also synthesizes recent developments in the field, highlighting the shift from traditional systems to advanced models like fine-tuned cascaded architectures and end-to-end solutions. In doing so, we surface the ongoing challenges, such as the need for realistic evaluation benchmarks, multilingual datasets, and long-context handling.