Ansh Chaurasia

Ansh Chaurasia

Researcher hopeful, AI generalist. Backpropogating on intelligent machines from novel attention mechanisms down to the silicon interposers.

Publications

You can also find my articles on my Google Scholar profile.

Journal Articles

MemorySim: An RTL-level, timing accurate simulator model for the Chisel ecosystem

Published in IEEE, 2025

We introduce MemorySim, a RTL-level memory simulator that strives to provide accurate timing simulations of memory systems while retaining correctness. The simulator is designed to be integrated into existing RTL-level simulations written in Chisel or Verilog. It’s also compatible with the Chisel / Chipyard ecosystem, allowing users to obtain highly accurate representations of performance and power through direct integration downstream leverage of simulation tools such as FireSim.

Download Paper

MANTA – Model Adapter Native generations that’s Affordable

Published in CVPR, 2024

The presiding model generation algorithms rely on simple, inflexible adapter selection to provide personalized results. We propose the model-adapter composition problem as a generalized problem to past work factoring in practical hardware and affordability constraints, and introduce MANTA as a new approach to the problem. Experiments on COCO 2014 validation show MANTA to be superior in image task diversity and quality at the cost of a modest drop in alignment. Our system achieves a 94% win rate in task diversity and a 80% task quality win rate versus the best known system, and demonstrates strong potential for direct use in synthetic data generation and the creative art domains.

Download Paper

Stylus: Automatic Adapter Selection for Diffusion Models

Published in Neurips, 2024

Beyond scaling base models with more data or parameters, fine-tuned adapters provide an alternative way to generate high fidelity, custom images at reduced costs. As such, adapters have been widely adopted by open-source communities, accumulating a database of over 100K adapters—most of which are highly customized with insufficient descriptions. To generate high quality images, this paper explores the problem of matching the prompt to a Stylus of relevant adapters, built on recent work that highlight the performance gains of composing adapters. We introduce Stylus, which efficiently selects and automatically composes task-specific adapters based on a prompt keywords. Stylus outlines a three-stage approach that first summarizes adapters with improved descriptions and embeddings, retrieves relevant adapters, and then further assembles adapters based on prompt keywords by checking how well they fit the prompt. To evaluate Stylus, we developed StylusDocs, a curated dataset featuring 75K adapters with pre-computed adapter embeddings. In our evaluation on popular Stable Diffusion checkpoints, Stylus achieves greater CLIP/FID Pareto efficiency and is twice as preferred, with humans and multimodal models as evaluators, over the base model.

Download Paper

Ball detection and tracking device, system and method

Published in USPTO, 2018

A ball detection and tracking system including one or more visual sensors and a detection and tracking agent that ranks a plurality of blob detection algorithms based on a detection metric and uses a selected base detection algorithm to identify one or more candidate blobs. Based on this, the agent is able to generate a track for the candidate blobs and assign one or more subsequent candidate blobs to a best ranked one of the tracks if the assignment satisfies a cost threshold.

Download Paper