Raghav Singhal

Flow Map Learning via Nongradient Vector Flow

Mark Goldstein, Anshuk Uppal, Raghav Singhal, Aahlad Puli et autres

Diffusion and flow-based models benefit from simple regression losses, but inference incurs significant overhead because sampling requires integration. Consistency models address this by directly learning the flow maps along the ODE trajectory, opening a design space between one-step and many-step approaches. However, …

Flow Map Learning via Nongradient Vector Flow

Mark Goldstein, Anshuk Uppal, Raghav Singhal, Aahlad Puli et autres

Diffusion and flow-based models benefit from simple regression losses, but inference incurs significant overhead because sampling requires integration. Consistency models address this by directly learning the flow maps along the ODE trajectory, opening a design space between one-step and many-step approaches. However, …

us, dk (code pays fourni par la source)

Imaging without Images: Using Artificial Intelligence for Direct Discovery of Spatial Signatures in the Absence of Image Reconstruction

Sumit Chopra, Raghav Singhal, Angela Tong, Rajesh Ranganath et autres

us (code pays fourni par la source)

A benchmark of expert-level academic questions to assess AI capabilities

Alice Gatti, Nathaniel Li, Adam Khoja, Ryan Kim et autres

Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve more than 90% accuracy on popular benchmarks such as Measuring Massive Multitask Language Understanding1, limiting informed …

us, id, gb, in, fr, ca, ua, bs, de (code pays fourni par la source)

A benchmark of expert-level academic questions to assess AI capabilities

Center for AI Safety, Long Phan, Dan Hendrycks, Haoran Zhao et autres

Center for AI Safety; Phan, Long; Gatti, Alice; Li, Nathaniel; Khoja, Adam; Kim, Ryan; Ren, Richard; Hausenloy, Jason; Zhang, Oliver; Mazeika, Mantas; Hendrycks, Dan; HLE Contributors Consortium; Han, Ziwen; Hu, Josephina; Zhang, Hugh; Zhang, Chen Bo Calvin; Shaaban, Mohamed; Ling, John; Shi, …

Rethinking Reasoning with MDLMs: Early Exits, Post-hoc Reasoning, and Beyond

Zachary Horvitz, Raghav Singhal, Hao Zou, Carles Domingo-Enrich et autres

The reasoning paradigm, where language models reason before answering, has enabled breakthroughs on tasks such as mathematical problem-solving. While current tooling for reasoning is built around next-token prediction trained models, recent works introduce an alternative choice: masked diffusion language models (MDLMs). MDLMs …

Analog and Temporary On-chip Memory for ANN Training and Inference

Shreyas Deshmukh, Shruti Landge, Raghav Singhal, Vivek Saraswat et autres

On-chip training at the edge becomes a primary requisite for real-time and security-sensitive artificial neural network (ANN) applications. In-memory computation (IMC) techniques have been proposed to facilitate data-intensive computational operations in ANNs. IMC with bidirectional multiply-accumulate (MAC) accelerates ANN on-chip training but …

Venous Blood Gas (VBG) Analysis Is as Safe and Equally Reliable as Arterial Blood Gas (ABG) Analysis in the Determination of Prognosis in Chronic Liver Disease Patients: A Study Conducted in a Tertiary Care Hospital in Uttar Pradesh, India

Manish Bansal, Raghav Singhal, Mayank Sharma, Chandra Prakash

Introduction Chronic liver disease (CLD) frequently causes systemic complications, including acid-base disturbances, significantly influencing patient prognosis. Arterial blood gas (ABG) analysis is traditionally utilized to monitor these disturbances, but presents procedural risks, especially in patients with coagulopathies, which is a well-known complication …

in (code pays fourni par la source)

Analog and Temporary On-chip Memory for ANN Training and Inference

Shreyas Deshmukh, Shruti Landge, Raghav Singhal, Vivek Saraswat et autres

On-chip training at the edge becomes a primary requisite for real-time and security-sensitive artificial neural network (ANN) applications. In-memory computation (IMC) techniques have been proposed to facilitate data-intensive computational operations in ANNs. IMC with bidirectional multiply-accumulate (MAC) accelerates ANN on-chip training but …

Regularization-based Framework for Quantization-, Fault- and Variability-Aware Training

Anmol Biswas, Raghav Singhal, Sivakumar Elangovan, Udayan Ganguly

Efficient inference is critical for deploying deep learning models on edge AI devices. Low-bit quantization (e.g., 3- and 4-bit) with fixed-point arithmetic improves efficiency, while low-power memory technologies like analog nonvolatile memory enable further gains. However, these methods introduce non-ideal hardware behavior, …

A General Framework for Inference-time Scaling and Steering of Diffusion Models

Raghav Singhal, Zachary Horvitz, Ryan Teehan, Mengye Ren et autres

Diffusion models produce impressive results in modalities ranging from images and video to protein design and text. However, generating samples with user-specified properties remains a challenge. Recent research proposes fine-tuning models to maximize rewards that capture desired properties, but these methods require …

Temporal and Spatial Reservoir Ensembling Techniques for Liquid State Machines

Anmol Biswas, Sharvari Ashok Medhe, Raghav Singhal, Udayan Ganguly

Reservoir computing (RC), is a class of computational methods such as Echo State Networks (ESN) and Liquid State Machines (LSM) describe a generic method to perform pattern recognition and temporal analysis with any non-linear system. This is enabled by Reservoir Computing being …