49 citations · 54 across the 11 of their papers we have counts for
11 papers
Deep sequence models tend to memorize geometrically; it is unclear why
Shahriar Noroozizadeh, Vaishnavh Nagarajan, Elan Rosenfeld +1
Deep sequence models are said to store atomic facts predominantly in the form of associative memory: a brute-force lookup of co-occurring entities. We identify a dramatically diffe…
Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities
Gheorghe Comanici, Eric Bieber, Mike Schaekermann +3431
In this report, we introduce the Gemini 2.X model family: Gemini 2.5 Pro and Gemini 2.5 Flash, as well as our earlier Gemini 2.0 Flash and Flash-Lite models. Gemini 2.5 Pro is our…
Structured Preconditioners in Adaptive Optimization: A Unified Analysis
Shuo Xie, Tianhao Wang, Sashank Reddi +2
We present a novel unified analysis for a broad class of adaptive optimization algorithms with structured (e.g., layerwise, diagonal, and kronecker-factored) preconditioners for bo…
Reasoning with Latent Thoughts: On the Power of Looped Transformers
Nikunj Saunshi, Nishanth Dikkala, Zhiyuan Li +2
Large language models have shown remarkable reasoning abilities and scaling laws suggest that large parameter count, especially along the depth axis, is the primary driver. In this…
LAuReL: Learned Augmented Residual Layer
Gaurav Menghani, Ravi Kumar, Sanjiv Kumar
One of the core pillars of efficient deep learning methods is architectural improvements such as the residual/skip connection, which has led to significantly better model convergen…
On the Role of Depth and Looping for In-Context Learning with Task Diversity
Khashayar Gatmiry, Nikunj Saunshi, Sashank J. Reddi +2
The intriguing in-context learning (ICL) abilities of deep Transformer models have lately garnered significant attention. By studying in-context linear regression on unimodal Gauss…
Can Looped Transformers Learn to Implement Multi-step Gradient Descent for In-context Learning?
Khashayar Gatmiry, Nikunj Saunshi, Sashank J. Reddi +2
The remarkable capability of Transformers to do reasoning and few-shot learning, without any fine-tuning, is widely conjectured to stem from their ability to implicitly simulate a…
On the Inductive Bias of Stacking Towards Improving Reasoning
Nikunj Saunshi, Stefani Karp, Shankar Krishnan +3
Given the increasing scale of model sizes, novel training strategies like gradual stacking [Gong et al., 2019, Reddi et al., 2023] have garnered interest. Stacking enables efficien…
Landscape-Aware Growing: The Power of a Little LAG
Stefani Karp, Nikunj Saunshi, Sobhan Miryoosefi +2
Recently, there has been increasing interest in efficient pretraining paradigms for training Transformer-based models. Several recent approaches use smaller models to initialize la…
SOAR: Improved Indexing for Approximate Nearest Neighbor Search
Philip Sun, David Simcha, Dave Dopson +2
This paper introduces SOAR: Spilling with Orthogonality-Amplified Residuals, a novel data indexing technique for approximate nearest neighbor (ANN) search. SOAR extends upon previo…
Efficient Stagewise Pretraining via Progressive Subnetworks
Abhishek Panigrahi, Nikunj Saunshi, Kaifeng Lyu +4
Recent developments in large language models have sparked interest in efficient pretraining methods. Stagewise training approaches to improve efficiency, like gradual stacking and…