Akhil Bhimaraju

Akhil Bhimaraju, Eliot W. Robson, Lav R. Varshney et al.

We consider the fractional influence maximization problem, i.e., identifying users on a social network to be incentivized with potentially partial discounts to maximize the influence on the network. The larger the discount given to a user, the higher the likelihood of its activation (adopting a new product or innovation), who then attempts to activate its neighboring users, causing a cascade effect of influence through the network. Our goal is to devise efficient algorithms that assign initial discounts to the network's users to maximize the total number of activated users at the end of the cascade, subject to a constraint on the total sum of discounts given. In general, the activation likelihood could be any non-decreasing function of the discount, whereas, our focus lies on the case when the activation likelihood is an affine function of the discount, potentially varying across different users. As this problem is shown to be NP-hard, we propose and analyze an efficient (1-1/e)-approximation algorithm. Furthermore, we run experiments on real-world social networks to show the performance and scalability of our method.

Avi Bagchi, Akhil Bhimaraju, Moulik Choraria et al.

Watermarking has emerged as a promising technique to track AI-generated content and differentiate it from authentic human creations. While prior work extensively studies watermarking for autoregressive large language models (LLMs) and image diffusion models, none address discrete diffusion language models, which are becoming popular due to their high inference throughput. In this paper, we introduce the first watermarking method for discrete diffusion models by applying the distribution-preserving Gumbel-max trick at every diffusion step and seeding the randomness with the sequence index to enable reliable detection. We experimentally demonstrate that our scheme is reliably detectable on state-of-the-art diffusion language models and analytically prove that it is distortion-free with an exponentially decaying probability of false detection in the token sequence length.

Moulik Choraria, Xinbo Wu, Akhil Bhimaraju et al. · amazon-science

The hyperscaling of data and parameter count in transformer models is yielding diminishing performance improvement, especially when weighed against training costs. Such plateauing underlines a growing need for more efficient finetuning and inference, without sacrificing performance. This is particularly pressing for multimodal learning, where the overhead of processing multimodal tokens alongside language data often limits the practical viability of these systems. In parallel, advances in representation learning and interpretability have deepened our understanding of how such models process and encode information. Notably, recent work has uncovered implicit cross-modal alignment in the deeper layers of large pretrained models. Interestingly, this aligns with our own observations that models naturally defer most cross-modal token interactions to deeper stages of computation. Building on this, we propose a simple modification. Instead of concatenation with the language prompt at the start, we insert multimodal tokens directly into the middle, allowing them to entirely bypass the early layers. Our results with diverse modalities: 1) LLaVA \& BLIP for vision, 2) LTU for audio, and 3) MoLCA for molecular data, indicate that our method reduces computational costs during both training and inference, while at the very least, preserving, if not surpassing the performance of existing baselines. Our work has important implications for scaling and composing pretrained models in a resource-efficient manner.

Max Hartman, Vidhata Jayaraman, Moulik Choraria et al.

Vision-language models (VLMs) achieve incredible performance across a wide range of tasks, but their large size makes inference costly. Recent work shows that selectively skipping VLM layers can improve efficiency with minimal performance loss or even performance improvements. However, this technique remains underused due to the limited understanding of when layer skipping is beneficial. In this paper, we develop a framework that uses information and learning theory to characterize the conditions under which layer skipping enhances efficiency without sacrificing performance. Motivated by these observations, we analyze the evolution of the VLM's hidden representations through the LLM backbone and show that layers with large redundancy as predicted by our framework coincide with those skipped by popular layer-skipping methods in practice, providing a unified theoretical scaffolding for multiple efficient inference techniques. Our experiments demonstrate that skipping such layers yields faster inference that preserves performance, and also show that applying skipping outside these conditions leads to model degradation.