arXiv subjects

Rajarshi Roy

Publications and source records attributed to Rajarshi Roy.

At least 37 records · Page 2Linked to original sources

arXiv2025

DETONATE: A Benchmark for Text-to-Image Alignment and Kernelized Direct Preference Optimization

Alignment is crucial for text-to-image (T2I) models to ensure that generated images faithfully capture user intent while maintaining safety and fairness. Direct Preference Optimization (DPO), prominent in large language models (LLMs), is extending its influence to T2I systems. This paper introduces DPO-Kernels for T2I models, a novel extension enhancing alignment across three dimensions: (i) Hybrid Loss, integrating embedding-based objectives with traditional probability-based loss for improved optimization; (ii) Kernelized Representations, employing Radial Basis Function (RBF), Polynomial, and Wavelet kernels for richer feature transformations and better separation between safe and unsafe inputs; and (iii) Divergence Selection, expanding beyond DPO's default Kullback-Leibler (KL) regularizer by incorporating Wasserstein and R'enyi divergences for enhanced stability and robustness. We introduce DETONATE, the first large-scale benchmark of its kind, comprising approximately 100K curated image pairs categorized as chosen and rejected. DETONATE encapsulates three axes of social bias and discrimination: Race, Gender, and Disability. Prompts are sourced from hate speech datasets, with images generated by leading T2I models including Stable Diffusion 3.5 Large, Stable Diffusion XL, and Midjourney. Additionally, we propose the Alignment Quality Index (AQI), a novel geometric measure quantifying latent-space separability of safe/unsafe image activations, revealing hidden vulnerabilities. Empirically, we demonstrate that DPO-Kernels maintain strong generalization bounds via Heavy-Tailed Self-Regularization (HT-SR). DETONATE and complete code are publicly released.

cs.CV↗

arXiv2025

Charge Ordering in out-of-plane Boron Doped Reduced Graphene Oxide

Symmetry-breaking phase transitions analogous to superconductivity (SC), charge ordering (CO) etc. in metal-intercalated graphene are favorable resulting from modified electronic and phonon band structures. Strong carrier-lattice interaction evolved from the out-of-plane soft vibrations with accumulation of charges at the out-of-plane region, can set a favorable environment for CO in graphene system. Here, we employ boron-doped reduced graphene oxide (BG) to acquire charge-ordered state above a transition temperature, T1~97.5 K. Signatures of this state are identified using ab-initio simulations and low temperature electrical transport measurements. The out-of-plane boron groups play a crucial role in reinforcing the electron-phonon coupling (EPC) allowing an ordered-state transition. Temperature-dependent Raman spectroscopy further supports the emergence of ordering. Key characterization techniques (X-ray diffraction, Raman spectra etc.) are used to quantify the EPC interaction and associated factors like tensile strain, boundary defects, etc. affecting charge ordering with doping. Additionally, we find interesting electric field dependency on the CO in this non-metallic, light-atom-doped chemically derived graphene.

cond-mat.mtrl-sci↗

arXiv2025

Learning Probabilistic Temporal Logic Specifications for Stochastic Systems

There has been substantial progress in the inference of formal behavioural specifications from sample trajectories, for example, using Linear Temporal Logic (LTL). However, these techniques cannot handle specifications that correctly characterise systems with stochastic behaviour, which occur commonly in reinforcement learning and formal verification. We consider the passive learning problem of inferring a Boolean combination of probabilistic LTL (PLTL) formulas from a set of Markov chains, classified as either positive or negative. We propose a novel learning algorithm that infers concise PLTL specifications, leveraging grammar-based enumeration, search heuristics, probabilistic model checking and Boolean set-cover procedures. We demonstrate the effectiveness of our algorithm in two use cases: learning from policies induced by RL algorithms and learning from variants of a probabilistic model. In both cases, our method automatically and efficiently extracts PLTL specifications that succinctly characterise the temporal differences between the policies or model variants.

cs.LO↗

arXiv2025

NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models

Decoder-only LLM-based embedding models are beginning to outperform BERT or T5-based embedding models in general-purpose text embedding tasks, including dense vector-based retrieval. In this work, we introduce NV-Embed, incorporating architectural designs, training procedures, and curated datasets to significantly enhance the performance of LLM as a versatile embedding model, while maintaining its simplicity and reproducibility. For model architecture, we propose a latent attention layer to obtain pooled embeddings, which consistently improves retrieval and downstream task accuracy compared to mean pooling or using the last token embedding from LLMs. To enhance representation learning, we remove the causal attention mask of LLMs during contrastive training. For training algorithm, we introduce a two-stage contrastive instruction-tuning method. It first applies contrastive training with instructions on retrieval datasets, utilizing in-batch negatives and curated hard negative examples. At stage-2, it blends various non-retrieval into instruction tuning, which not only enhances non-retrieval task accuracy but also improves retrieval performance. For training data, we utilize the hard-negative mining, synthetic data generation and existing public available datasets to boost the performance of embedding model. By combining these techniques, our NV-Embed-v1 and NV-Embed-v2 models obtained the No.1 position on the MTEB leaderboard (as of May 24 and August 30, 2024, respectively) across 56 tasks, demonstrating the sustained effectiveness of the proposed methods over time. It also achieved the highest scores in the Long Doc section and the second-highest scores in the QA section of the AIR Benchmark, which covers a range of out-of-domain information retrieval topics beyond those in MTEB. We further provide the analysis of model compression techniques for generalist embedding models.

cs.CL↗

arXiv2025

What is Formal Verification without Specifications? A Survey on mining LTL Specifications

Virtually all verification techniques using formal methods rely on the availability of a formal specification, which describes the design requirements precisely. However, formulating specifications remains a manual task that is notoriously challenging and error-prone. To address this bottleneck in formal verification, recent research has thus focussed on automatically generating specifications for formal verification from examples of (desired and undesired) system behavior. In this survey, we list and compare recent advances in mining specifications in Linear Temporal Logic (LTL), the de facto standard specification language for reactive systems. Several approaches have been designed for learning LTL formulas, which address different aspects and settings of specification design. Moreover, the approaches rely on a diverse range of techniques such as constraint solving, neural network training, enumerative search, etc. We survey the current state-of-the-art techniques and compare them for the convenience of the formal methods practitioners.

cs.FL↗

arXiv2025

DPO Kernels: A Semantically-Aware, Kernel-Enhanced, and Divergence-Rich Paradigm for Direct Preference Optimization

The rapid rise of large language models (LLMs) has unlocked many applications but also underscores the challenge of aligning them with diverse values and preferences. Direct Preference Optimization (DPO) is central to alignment but constrained by fixed divergences and limited feature transformations. We propose DPO-Kernels, which integrates kernel methods to address these issues through four key contributions: (i) Kernelized Representations with polynomial, RBF, Mahalanobis, and spectral kernels for richer transformations, plus a hybrid loss combining embedding-based and probability-based objectives; (ii) Divergence Alternatives (Jensen-Shannon, Hellinger, Renyi, Bhattacharyya, Wasserstein, and f-divergences) for greater stability; (iii) Data-Driven Selection metrics that automatically choose the best kernel-divergence pair; and (iv) a Hierarchical Mixture of Kernels for both local precision and global modeling. Evaluations on 12 datasets demonstrate state-of-the-art performance in factuality, safety, reasoning, and instruction following. Grounded in Heavy-Tailed Self-Regularization, DPO-Kernels maintains robust generalization for LLMs, offering a comprehensive resource for further alignment research.

cs.LG↗

arXiv2024

ChatQA: Surpassing GPT-4 on Conversational QA and RAG

In this work, we introduce ChatQA, a suite of models that outperform GPT-4 on retrieval-augmented generation (RAG) and conversational question answering (QA). To enhance generation, we propose a two-stage instruction tuning method that significantly boosts the performance of RAG. For effective retrieval, we introduce a dense retriever optimized for conversational QA, which yields results comparable to the alternative state-of-the-art query rewriting models, while substantially reducing deployment costs. We also present the ChatRAG Bench, which encompasses ten datasets covering comprehensive evaluations on RAG, table-related QA, arithmetic calculations, and scenarios involving unanswerable questions. Our ChatQA-1.0-70B (score: 54.14), built on Llama2, a weaker foundation model than GPT-4, can slightly outperform GPT-4-0613 (score: 53.90) and GPT-4-Turbo-2024-04-09 (score: 54.03) on the ChatRAG Bench, without relying on any synthetic data from OpenAI GPT models. Notably, the Llama3-ChatQA-1.5-70B model surpasses the accuracy of GPT-4-Turbo-2024-04-09, achieving a 4.4% improvement. To advance research in this field, we open-sourced the model weights, instruction tuning data, ChatRAG Bench, and retriever for the community: https://chatqa-project.github.io/.

cs.CL↗

arXiv2024

The Complexity of Learning Temporal Properties

We consider the problem of learning temporal logic formulas from examples of system behavior. Learning temporal properties has crystallized as an effective mean to explain complex temporal behaviors. Several efficient algorithms have been designed for learning temporal formulas. However, the theoretical understanding of the complexity of the learning decision problems remains largely unexplored. To address this, we study the complexity of the passive learning problems of three prominent temporal logics, Linear Temporal Logic (LTL), Computation Tree Logic (CTL) and Alternating-time Temporal Logic (ATL) and several of their fragments. We show that learning formulas using an unbounded amount of occurrences of binary operators is NP-complete for all of these logics. On the other hand, when investigating the complexity of learning formulas with bounded amount of occurrences of binary operators, we exhibit discrepancies between the complexity of learning LTL, CTL and ATL formulas (with a varying number of agents).

cs.LO↗

arXiv2024

Nemotron-4 340B Technical Report

We release the Nemotron-4 340B model family, including Nemotron-4-340B-Base, Nemotron-4-340B-Instruct, and Nemotron-4-340B-Reward. Our models are open access under the NVIDIA Open Model License Agreement, a permissive model license that allows distribution, modification, and use of the models and its outputs. These models perform competitively to open access models on a wide range of evaluation benchmarks, and were sized to fit on a single DGX H100 with 8 GPUs when deployed in FP8 precision. We believe that the community can benefit from these models in various research studies and commercial applications, especially for generating synthetic data to train smaller language models. Notably, over 98% of data used in our model alignment process is synthetically generated, showcasing the effectiveness of these models in generating synthetic data. To further support open research and facilitate model development, we are also open-sourcing the synthetic data generation pipeline used in our model alignment process.

cs.CL↗

arXiv2024

A High-Fidelity Simulation Framework for Grasping Stability Analysis in Human Casualty Manipulation

Recently, there has been a growing interest in rescue robots due to their vital role in addressing emergency scenarios and providing crucial support in challenging or hazardous situations where human intervention is difficult. However, very few of these robots are capable of actively engaging with humans and undertaking physical manipulation tasks. This limitation is largely attributed to the absence of tools that can realistically simulate physical interactions, especially the contact mechanisms between a robotic gripper and a human body. In this letter, we aim to address key limitations in current developments towards robotic casualty manipulation. Firstly, we present an integrative simulation framework for casualty manipulation. We adapt a finite element method (FEM) tool into the grasping and manipulation scenario, and the developed framework can provide accurate biomechanical reactions resulting from manipulation. Secondly, we conduct a detailed assessment of grasping stability during casualty grasping and manipulation simulations. To validate the necessity and superior performance of the proposed high-fidelity simulation framework, we conducted a qualitative and quantitative comparison of grasping stability analyses between the proposed framework and the state-of-the-art multi-body physics simulations. Through these efforts, we have taken the first step towards a feasible solution for robotic casualty manipulation.

cs.RO↗

arXiv2024

Learning Branching-Time Properties in CTL and ATL via Constraint Solving

We address the problem of learning temporal properties from the branching-time behavior of systems. Existing research in this field has mostly focused on learning linear temporal properties specified using popular logics, such as Linear Temporal Logic (LTL) and Signal Temporal Logic (STL). Branching-time logics such as Computation Tree Logic (CTL) and Alternating-time Temporal Logic (ATL), despite being extensively used in specifying and verifying distributed and multi-agent systems, have not received adequate attention. Thus, in this paper, we investigate the problem of learning CTL and ATL formulas from examples of system behavior. As input to the learning problems, we rely on the typical representations of branching behavior as Kripke structures and concurrent game structures, respectively. Given a sample of structures, we learn concise formulas by encoding the learning problem into a satisfiability problem, most notably by symbolically encoding both the search for prospective formulas and their fixed-point based model checking algorithms. We also study the decision problem of checking the existence of prospective ATL formulas for a given sample. We implement our algorithms in an Python prototype and have evaluated them to extract several common CTL and ATL formulas used in practical applications.

cs.LO↗

arXiv2024

CircuitVAE: Efficient and Scalable Latent Circuit Optimization

Automatically designing fast and space-efficient digital circuits is challenging because circuits are discrete, must exactly implement the desired logic, and are costly to simulate. We address these challenges with CircuitVAE, a search algorithm that embeds computation graphs in a continuous space and optimizes a learned surrogate of physical simulation by gradient descent. By carefully controlling overfitting of the simulation surrogate and ensuring diverse exploration, our algorithm is highly sample-efficient, yet gracefully scales to large problem instances and high sample budgets. We test CircuitVAE by designing binary adders across a large range of sizes, IO timing constraints, and sample budgets. Our method excels at designing large circuits, where other algorithms struggle: compared to reinforcement learning and genetic algorithms, CircuitVAE typically finds 64-bit adders which are smaller and faster using less than half the sample budget. We also find CircuitVAE can design state-of-the-art adders in a real-world chip, demonstrating that our method can outperform commercial tools in a realistic setting.

cs.LG↗

arXiv2024

Analyzing Robustness of Angluin's L$^*$ Algorithm in Presence of Noise

Angluin's L$^*$ algorithm learns the minimal deterministic finite automaton (DFA) of a regular language using membership and equivalence queries. Its probabilistic approximatively correct (PAC) version substitutes an equivalence query by numerous random membership queries to get a high level confidence to the answer. Thus it can be applied to any kind of device and may be viewed as an algorithm for synthesizing an automaton abstracting the behavior of the device based on observations. Here we are interested on how Angluin's PAC learning algorithm behaves for devices which are obtained from a DFA by introducing some noise. More precisely we study whether Angluin's algorithm reduces the noise and produces a DFA closer to the original one than the noisy device. We propose several ways to introduce the noise: (1) the noisy device inverts the classification of words w.r.t. the DFA with a small probability, (2) the noisy device modifies with a small probability the letters of the word before asking its classification w.r.t. the DFA, (3) the noisy device combines the classification of a word w.r.t. the DFA and its classification w.r.t. a counter automaton, and (4) the noisy DFA is obtained by a random process from two DFA such that the language of the first one is included in the second one. Then when a word is accepted (resp. rejected) by the first (resp. second) one, it is also accepted (resp. rejected) and in the remaining cases, it is accepted with probability 0.5. Our main experimental contributions consist in showing that: (1) Angluin's algorithm behaves well whenever the noisy device is produced by a random process, (2) but poorly with a structured noise, and, that (3) is able to eliminate pathological behaviours specified in a regular way. Theoretically, we show that randomness almost surely yields systems with non-recursively enumerable languages.

cs.FL↗

arXiv2023

Learning Temporal Properties is NP-hard

We investigate the complexity of LTL learning, which consists in deciding given a finite set of positive ultimately periodic words, a finite set of negative ultimately periodic words, and a bound B given in unary, if there is an LTL-formula of size less than or equal to B that all positive words satisfy and that all negative violate. We prove that this decision problem is NP-hard. We then use this result to show that CTL learning is also NP-hard. CTL learning is similar to LTL learning except that words are replaced by finite Kripke structures and we look for the existence of CTL formulae.

cs.LO↗

arXiv2023

Synthesizing Efficiently Monitorable Formulas in Metric Temporal Logic

In runtime verification, manually formalizing a specification for monitoring system executions is a tedious and error-prone process. To address this issue, we consider the problem of automatically synthesizing formal specifications from system executions. To demonstrate our approach, we consider the popular specification language Metric Temporal Logic (MTL), which is particularly tailored towards specifying temporal properties for cyber-physical systems (CPS). Most of the classical approaches for synthesizing temporal logic formulas aim at minimizing the size of the formula. However, for efficiency in monitoring, along with the size, the amount of "lookahead" required for the specification becomes relevant, especially for safety-critical applications. We formalize this notion and devise a learning algorithm that synthesizes concise formulas having bounded lookahead. To do so, our algorithm reduces the synthesis task to a series of satisfiability problems in Linear Real Arithmetic (LRA) and generates MTL formulas from their satisfying assignments. The reduction uses a novel encoding of a popular MTL monitoring procedure using LRA. Finally, we implement our algorithm in a tool called TEAL and demonstrate its ability to synthesize efficiently monitorable MTL formulas in a CPS application.

cs.AI↗

arXiv2023

Robust Computation Tree Logic

It is widely accepted that every system should be robust in that ``small'' violations of environment assumptions should lead to ``small'' violations of system guarantees, but it is less clear how to make this intuition mathematically precise. While significant efforts have been devoted to providing notions of robustness for Linear Temporal Logic (LTL), branching-time logics, such as Computation Tree Logic (CTL) and CTL*, have received less attention in this regard. To address this shortcoming, we develop ``robust'' extensions of CTL and CTL*, which we name robust CTL (rCTL) and robust CTL* (rCTL*). Both extensions are syntactically similar to their parent logics but employ multi-valued semantics to distinguish between ``large'' and ``small'' violations of the specification. We show that the multi-valued semantics of rCTL make it more expressive than CTL, while rCTL* is as expressive as CTL*. Moreover, we show that the model checking problem, the satisfiability problem, and the synthesis problem for rCTL and rCTL* have the same asymptotic complexity as their non-robust counterparts, implying that robustness can be added to branching-time logics for free.

cs.LO↗

arXiv2023

Inferring Properties in Computation Tree Logic

We consider the problem of automatically inferring specifications in the branching-time logic, Computation Tree Logic (CTL), from a given system. Designing functional and usable specifications has always been one of the biggest challenges of formal methods. While in recent years, works have focused on automatically designing specifications in linear-time logics such as Linear Temporal Logic (LTL) and Signal Temporal Logic (STL), little attention has been given to branching-time logics despite its popularity in formal methods. We intend to infer concise (thus, interpretable) CTL formulas from a given finite state model of the system in consideration. However, inferring specification only from the given model (and, in general, from only positive examples) is an ill-posed problem. As a result, we infer a CTL formula that, along with being concise, is also language-minimal, meaning that it is rather specific to the given model. We design a counter-example guided algorithm to infer a concise and language-minimal CTL formula via the generation of undesirable models. In the process, we also develop, for the first time, a passive learning algorithm to infer CTL formulas from a set of desirable and undesirable Kripke structures. The passive learning algorithm involves encoding a popular CTL model-checking procedure in the Boolean Satisfiability problem.

cs.LO↗

arXiv2023

GraPhSyM: Graph Physical Synthesis Model

In this work, we introduce GraPhSyM, a Graph Attention Network (GATv2) model for fast and accurate estimation of post-physical synthesis circuit delay and area metrics from pre-physical synthesis circuit netlists. Once trained, GraPhSyM provides accurate visibility of final design metrics to early EDA stages, such as logic synthesis, without running the slow physical synthesis flow, enabling global co-optimization across stages. Additionally, the swift and precise feedback provided by GraPhSyM is instrumental for machine-learning-based EDA optimization frameworks. Given a gate-level netlist of a circuit represented as a graph, GraPhSyM utilizes graph structure, connectivity, and electrical property features to predict the impact of physical synthesis transformations such as buffer insertion and gate sizing. When trained on a dataset of 6000 prefix adder designs synthesized at an aggressive delay target, GraPhSyM can accurately predict the post-synthesis delay (98.3%) and area (96.1%) metrics of unseen adders with a fast 0.22s inference time. Furthermore, we illustrate the compositionality of GraPhSyM by employing the model trained on a fixed delay target to accurately anticipate post-synthesis metrics at a variety of unseen delay targets. Lastly, we report promising generalization capabilities of the GraPhSyM model when it is evaluated on circuits different from the adders it was exclusively trained on. The results show the potential for GraPhSyM to serve as a powerful tool for advanced optimization techniques and as an oracle for EDA machine learning frameworks.

cs.LG↗