DISC: Dynamic Decomposition Improves LLM Inference Scaling (SSI-FM)

/in /by

Inference scaling methods often rely on decomposing problems into steps, followed by sampling and selecting the best next steps. However, these steps and their sizes are typically fixed or depend on domain knowledge. We propose dynamic decomposition, a method that adaptively and automatically breaks down solution and reasoning traces into manageable steps during inference. By allocating compute more effectively, particularly by subdividing challenging steps and sampling them more frequently, dynamic decomposition significantly enhances inference efficiency. Experiments on benchmarks such as APPS, MATH, and LiveCodeBench demonstrate that dynamic decomposition outperforms static approaches, including token-level, sentence-level, and single-step decompositions. These findings highlight the potential of dynamic decomposition to improve a wide range of inference scaling techniques.

Read more

SFS: Smarter Code Space Search improves LLM Inference Scaling

/in /by

We frame code generation as a black-box optimization problem within the code space and demonstrate how optimization-inspired techniques can enhance inference scaling. Based on this perspective, we propose SCATTERED FOREST SEARCH (SFS), a novel approach that improves solution diversity and better exploits feedback during evolutionary search. Our theoretical analysis illustrates how these methods help avoid local optima during optimization, leading to more efficient exploration. Extensive experiments on HumanEval, MBPP, APPS, CodeContests, and Leetcode reveal significant performance gains. For instance, our method achieves a pass@1 rate of 67.1% on HumanEval+ and 87.2% on HumanEval with GPT-3.5, marking improvements of 8.6% and 4.3% over the state-of-the-art, while also halving the iterations needed to find the correct solution. Furthermore, our approach scales more efficiently than existing search techniques, including tree search, line search, and repeated sampling.

Read more

Humanizing the Machine: Proxy Attacks to Mislead LLM Detectors

/in /by

The advent of large language models (LLMs) has revolutionized the field of text generation, producing outputs that closely mimic human-like writing. Although academic and industrial institutions have developed detectors to prevent the malicious usage of LLM-generated texts, other research has doubt about the robustness of these systems. To stress test these detectors, we introduce a humanized proxy-attack (HUMPA) strategy that effortlessly compromises LLMs, causing them to produce outputs that align with human-written text and mislead detection systems. Our method attacks the source model by leveraging a reinforcement learning (RL) fine-tuned humanized small language model (SLM) in the decoding phase. Through an in-depth analysis, we demonstrate that our attack strategy is capable of generating responses that are indistinguishable to detectors, preventing them from differentiating between machine-generated and human-written text. We conduct systematic evaluations on extensive datasets using proxy-attacked open-source models, including Llama2-13B, Llama3-70B, and Mixtral-8×7B in both white- and black-box settings. Our findings show that the proxy-attack strategy effectively deceives the leading detectors, resulting in an average AUROC drop of 70.4% across multiple datasets, with a maximum drop of 95.0% on a single dataset. Furthermore, in cross-discipline scenarios, our strategy also bypasses these detectors, leading to a significant relative decrease of up to 90.9%, while in cross-language scenario, the drop reaches 91.3%. Despite our proxy-attack strategy successfully bypassing the detectors with such significant relative drops, we find that the generation quality of the attacked models remains preserved, even within a modest utility budget, when compared to the text produced by the original, unattacked source model.

Read more

F-Fidelity: A Robust Framework for Faithful-NESS Evaluation in Explainable AI

/in /by

Recent research has developed a number of eXplainable AI (XAI) techniques, such as gradient-based approaches, input perturbation-base methods, and black-box explanation methods. While these XAI techniques can extract meaningful insights from deep learning models, how to properly evaluate them remains an open problem. The most widely used approach is to perturb or even remove what the XAI method considers to be the most important features in an input and observe the changes in the output prediction. This approach, although straightforward, suffers the Out-of-Distribution (OOD) problem as the perturbed samples may no longer follow the original data distribution. A recent method RemOve And Retrain (ROAR) solves the OOD issue by retraining the model with perturbed samples guided by explanations. However, using the model retrained based on XAI methods to evaluate these explainers may cause information leakage and thus lead to unfair comparisons. We propose Fine-tuned Fidelity (F-Fidelity), a robust evaluation framework for XAI, which utilizes i) an explanation-agnostic fine-tuning strategy, thus mitigating the information leakage issue, and ii) a random masking operation that ensures that the removal step does not generate an OOD input. We also design controlled experiments with state-of-the-art (SOTA) explainers and their degraded version to verify the correctness of our framework. We conduct experiments on multiple data modalities, such as images, time series, and natural language. The results demonstrate that F-Fidelity significantly improves upon prior evaluation metrics in recovering the ground-truth ranking of the explainers. Furthermore, we show both theoretically and empirically that, given a faithful explainer, F-Fidelity metric can be used to compute the sparsity of influential input components, i.e., to extract the true explanation size.

Read more

DISC: Dynamic Decomposition Improves LLM Inference Scaling (DL4C)

/in /by

Inference scaling methods often rely on decomposing problems into steps, followed by sampling and selecting the best next steps. However, these steps and their sizes are typically fixed or depend on domain knowledge. We propose dynamic decomposition, a method that adaptively and automatically breaks down solution and reasoning traces into manageable steps during inference. By allocating compute more effectively—particularly by subdividing challenging steps and sampling them more frequently—dynamic decomposition significantly enhances inference efficiency. Experiments on benchmarks such as APPS, MATH, and LiveCodeBench demonstrate that dynamic decomposition outperforms static approaches, including token-level, sentence-level, and single-step decompositions. These findings highlight the potential of dynamic decomposition to improve a wide range of inference scaling techniques.

Read more

Chain-of-region: Visual Language Models Need Details for Diagram Analysis

/in /by

Visual Language Models (VLMs) like GPT-4V have broadened the scope of LLM applications, yet they face significant challenges in accurately processing visual details, particularly in scientific diagrams. This paper explores the necessity of meticulous visual detail collection and region decomposition for enhancing the performance of VLMs in scientific diagram analysis. We propose a novel approach that combines traditional computer vision techniques with VLMs to systematically decompose diagrams into discernible visual elements and aggregate essential metadata. Our method employs techniques in OpenCV library to identify and label regions, followed by a refinement process using shape detection and region merging algorithms, which are particularly suited to the structured nature of scientific diagrams. This strategy not only improves the granularity and accuracy of visual information processing but also extends the capabilities of VLMs beyond their current limitations. We validate our approach through a series of experiments that demonstrate enhanced performance in diagram analysis tasks, setting a new standard for integrating visual and language processing in a multimodal context.

Read more

Progressive Token Length Scaling in Transformer Encoders for Efficient Universal Segmentation

/in /by

A powerful architecture for universal segmentation relies on transformers that encode multi-scale image features and decode object queries into mask predictions. With efficiency being a high priority for scaling such models, we observed that the state-of-the-art method Mask2Former uses >50% of its compute only on the transformer encoder. This is due to the retention of a full-length token-level representation of all backbone feature scales at each encoder layer. With this observation, we propose a strategy termed PROgressive Token Length SCALing for Efficient transformer encoders (PRO-SCALE) that can be plugged-in to the Mask2Former segmentation architecture to significantly reduce the computational cost. The underlying principle of PRO-SCALE is: progressively scale the length of the tokens with the layers of the encoder. This allows PRO-SCALE to reduce computations by a large margin with minimal sacrifice in performance (?52% encoder and ? 27% overall GFLOPs reduction with no drop in performance on COCO dataset). Experiments conducted on public benchmarks demonstrates PRO-SCALE’s flexibility in architectural configurations, and exhibits potential for extension beyond the settings of segmentation tasks to encompass object detection. Code available here: https://github.com/abhishekaich27/proscale-pytorch

Read more

A Quantum Variational Autoencoder Utilizing Regularized Mixed-state Latent Representations

/in /by

A major challenge in near-term quantum computing is its application to large real-world datasets due to scarce quantum hardware resources. One approach to enabling tractable quantum models for such datasets involves finding low-dimensional representations that preserve essential information for downstream analysis. Inclassical machine learning, variational autoencoders (VAEs) facilitate efficient data compression, representationlearning for subsequent tasks, and novel data generation. However, no quantum model has been proposed thatexactly captures all of these features for direct application to quantum data on quantum computers. Some existingquantum models for data compression lack regularization of latent representations, thus preventing direct use forgeneration and control of generalization. Others are hybrid models with only some internal quantum components,impeding direct training on quantum data. To address this, we present a fully quantum framework, ?-QVAE,which encompasses all the capabilities of classical VAEs and can be directly applied to map both classicaland quantum data to a lower-dimensional space, while effectively reconstructing much of the original statefrom it. Our model utilizes regularized mixed states to attain optimal latent representations. It accommodatesvarious divergences for reconstruction and regularization. Furthermore, by accommodating mixed states at everystage, it can utilize the full data density matrix and allow for a training objective defined on probabilisticmixtures of input data. Doing so, in turn, makes efficient optimization possible and has potential implications forprivate and federated learning. In addition to exploring the theoretical properties of ?-QVAE, we demonstrateits performance on representative genomics and synthetic data. Our results indicate that ?-QVAE consistentlylearns representations that better utilize the capacity of the latent space and exhibits similar or better performancecompared with matched classical models.

Read more

TSLA: Unified Time Series and Language Model

/in /by

Real-world time series data often require analysis or interpretation from domain experts. Some tasks, like time series question answering, involve both time series and natural language questions, posing challenges for single-modality language models to understand their interaction. To this end, we present TSLA (Time Series Language Model), a framework designed to enhance the language model with the understanding of time series data for multi-modality tasks. TSLA comprises three key components. (1) Time Series Tokenizer learns how to represent time series data into discrete tokens, making it more manageable for language models. (2) Joint (Pre-)Training on task-agnostic time series and text data integrates time series tokens and text tokens to model the interplay between time series and language concepts. (3) Multi-task Instruction Tuning fine-tunes the pretrained TSLA for various downstream tasks relevant to user interests. For evaluation, we applied TSLA to time series data from human motions on four tasks: time series captioning, time series question answering, text-based time series synthesis, and text-based time series continuation. The results demonstrate TSLA’s effectiveness in handling multiple time series analysis tasks, pointing the way for future research endeavors.

Read more

Graph Neural Networks, Explained: Our Role in the Future of AI

/in /by

NEC Laboratories America (NECLA) is advancing the frontier of Graph Neural Networks (GNNs), a transformative AI technology that processes complex, interconnected data. Through innovations like PTDNet for robust learning, novel frameworks for explainability, StrGNN for anomaly detection in dynamic graphs, and GERDQ for calibration with out-of-distribution nodes, NECLA is addressing critical challenges in GNN development. These breakthroughs have real-world implications in fields such as cybersecurity, bioinformatics, and recommendation systems, positioning NECLA as a leader in the evolution of graph-based AI.

Read more