外部记忆

This paper presents a novel approach to address the challenge of online sequence learning for decision making under uncertainty in non-stationary, partially observable environments. The proposed algorithm, Distributed Hebbian Temporal Memory (DHTM), is based on the factor graph formalism and a multi-component neuron model. DHTM aims to capture sequential data relationships and make cumulative predictions about future observations, forming Successor Features (SFs). Inspired by neurophysiological models of the neocortex, the algorithm uses distributed representations, sparse transition matrices, and local Hebbian-like learning rules to overcome the instability and slow learning of traditional temporal memory algorithms such as RNN and HMM. Experimental results show that DHTM outperforms LSTM, RWKV and a biologically inspired HMM-like algorithm, CSCG, on non-stationary data sets. Our results suggest that DHTM is a promising approach to address the challenges of online sequence learning and planning in dynamic environments.

Prior methods propose to offset the escalating costs of modern foundation models by dropping specific parts of their contexts with hand-designed rules, while attempting to preserve their original performance. We overcome this trade-off with Neural Attention Memory Models (NAMMs), introducing a learned network for memory management that improves both the performance and efficiency of transformers. We evolve NAMMs atop pre-trained transformers to provide different latent contexts focusing on the most relevant information for individual layers and attention heads. NAMMs are universally applicable to any model using self-attention as they condition exclusively on the values in the produced attention matrices. Learning NAMMs on a small set of problems, we achieve substantial performance improvements across multiple long-context benchmarks while cutting the model's input contexts up to a fraction of the original sizes. We show the generality of our conditioning enables zero-shot transfer of NAMMs trained only on language to entirely new transformer architectures even across input modalities, with their benefits carrying over to vision and reinforcement learning.

In this paper, we consider the problem of long-term point tracking, which requires consistent identification of points across multiple frames in a video, despite changes in appearance, lighting, perspective, and occlusions. We target online tracking on a frame-by-frame basis, making it suitable for real-world, streaming scenarios. Specifically, we introduce Track-On, a simple transformer-based model designed for online long-term point tracking. Unlike prior methods that depend on full temporal modeling, our model processes video frames causally without access to future frames, leveraging two memory modules —spatial memory and context memory— to capture temporal information and maintain reliable point tracking over long time horizons. At inference time, it employs patch classification and refinement to identify correspondences and track points with high accuracy. Through extensive experiments, we demonstrate that Track-On sets a new state-of-the-art for online models and delivers superior or competitive results compared to offline approaches on seven datasets, including the TAP-Vid benchmark. Our method offers a robust and scalable solution for real-time tracking in diverse applications. Project page: https://kuis-ai.github.io/track_on

Large language models (LLMs) have revolutionized numerous applications, yet their deployment remains challenged by memory constraints on local devices. While scaling laws have enhanced LLM capabilities, the primary bottleneck has shifted from $\textit{capability}$ to $\textit{availability}$, emphasizing the need for efficient memory management. Traditional compression methods, such as quantization, often require predefined compression ratios and separate compression processes for each setting, complicating deployment in variable memory environments. In this paper, we introduce $\textbf{BitStack}$, a novel, training-free weight compression approach that enables megabyte-level trade-offs between memory usage and model performance. By leveraging weight decomposition, BitStack can dynamically adjust the model size with minimal transmission between running memory and storage devices. Our approach iteratively decomposes weight matrices while considering the significance of each parameter, resulting in an approximately 1-bit per parameter residual block in each decomposition iteration. These blocks are sorted and stacked in storage as basic transmission units, with different quantities loaded based on current memory availability. Extensive experiments across a wide range of tasks demonstrate that, despite offering fine-grained size control, BitStack consistently matches or surpasses strong quantization baselines, particularly at extreme compression ratios. To the best of our knowledge, this is the first decomposition-based method that effectively bridges the gap to practical compression techniques like quantization. Code is available at https://github.com/xinghaow99/BitStack.

We propose Adam-mini, an optimizer that achieves on-par or better performance than AdamW with $50$% less memory footprint. Adam-mini reduces memory by cutting down the learning rate resources in Adam (i.e., $1/\sqrt{v}$). By delving into the Hessian structure of neural nets, we find Adam’s $v$ might not function at its full potential as effectively as we expected. We find that $\geq 99.9$% of these learning rates in $v$ could be harmlessly removed if we (1) carefully partition the parameters into blocks following our proposed principle on Hessian structure; (2) assign a single but good learning rate to each parameter block. We then provide one simple way to find good learning rates and propose Adam-mini. Empirically, we verify that Adam-mini performs on par or better than AdamW on various language models sized from 39M to 13B for pre-training, supervised fine-tuning, and RLHF. The reduced memory footprint of Adam-mini also alleviates communication overheads among GPUs, thereby increasing throughput. For instance, Adam-mini achieves $49.6$% higher throughput than AdamW when pre-training Llama 2-7B on $2\times$ A800-80GB GPUs, which saves 33% wall-clock time for pre-training.

We present 3D Spatial MultiModal Memory (M3), a multimodal memory system designed to retain information about medium-sized static scenes through video sources for visual perception. By integrating 3D Gaussian Splatting techniques with foundation models, M3 builds a multimodal memory capable of rendering feature representations across granularities, encompassing a wide range of knowledge. In our exploration, we identify two key challenges in previous works on feature splatting: (1) computational constraints in storing high-dimensional features for each Gaussian primitive, and (2) misalignment or information loss between distilled features and foundation model features. To address these challenges, we propose M3 with key components of principal scene components and Gaussian memory attention, enabling efficient training and inference. To validate M3, we conduct comprehensive quantitative evaluations of feature similarity and downstream tasks, as well as qualitative visualizations to highlight the pixel trace of Gaussian memory attention. Our approach encompasses a diverse range of foundation models, including vision-language models (VLMs), perception models, and large multimodal and language models (LMMs/LLMs). Furthermore, to demonstrate real-world applicability, we deploy M3’s feature field in indoor scenes on a quadruped robot. Notably, we claim that M3 is the first work to address the core compression challenges in 3D feature distillation.

Contextual semantic information plays a pivotal role in the brain's visual interpretation of the surrounding environment. When processing visual information, electrical signals within synapses facilitate the dynamic activation and deactivation of synaptic connections, guided by the contextual semantic information associated with different objects. In the realm of Artificial Intelligence (AI), neural networks have emerged as powerful tools to emulate complex signaling systems, enabling tasks such as classification and segmentation by understanding visual information. However, conventional neural networks have limitations in simulating the conditional activation and deactivation of synapses, collectively known as the connectome, a comprehensive map of neural connections in the brain. Additionally, the pixel-wise inference mechanism of conventional neural networks failed to account for the explicit utilization of contextual semantic information in the prediction process. To overcome these limitations, we developed a novel neural network, dubbed the Shape Memory Network (SMN), which excels in two key areas: (1) faithfully emulating the intricate mechanism of the brain's connectome, and (2) explicitly incorporating contextual semantic information during the inference process. The SMN memorizes the structure suitable for contextual semantic information and leverages this structure at the inference phase. The structural transformation emulates the conditional activation and deactivation of synaptic connections within the connectome. Rigorous experimentation carried out across a range of semantic segmentation benchmarks demonstrated the outstanding performance of the SMN, highlighting its superiority and effectiveness. Furthermore, our pioneering network on connectome emulation reveals the immense potential of the SMN for next-generation neural networks.

Memory Mosaics are networks of associative memories working in concert to achieve a prediction task of interest. Like transformers, memory mosaics possess compositional capabilities and in-context learning capabilities. Unlike transformers, memory mosaics achieve these capabilities in comparatively transparent way (“predictive disentanglement”). We illustrate these capabilities on a toy example and also show that memory mosaics perform as well or better than transformers on medium-scale language modeling tasks.

Significant advances have been made in developing general-purpose embodied AI in environments like Minecraft through the adoption of LLM-augmented hierarchical approaches. While these approaches, which combine high-level planners with low-level controllers, show promise, low-level controllers frequently become performance bottlenecks due to repeated failures. In this paper, we argue that the primary cause of failure in many low-level controllers is the absence of an episodic memory system. To address this, we introduce MrSteve (Memory Recall Steve), a novel low-level controller equipped with Place Event Memory (PEM), a form of episodic memory that captures what, where, and when information from episodes. This directly addresses the main limitation of the popular low-level controller, Steve-1. Unlike previous models that rely on short-term memory, PEM organizes spatial and event-based data, enabling efficient recall and navigation in long-horizon tasks. Additionally, we propose an Exploration Strategy and a Memory-Augmented Task Solving Framework, allowing agents to alternate between exploration and task-solving based on recalled events. Our approach significantly improves task-solving and exploration efficiency compared to existing methods. We will release our code and demos on the project page: https://sites.google.com/view/mr-steve.

Enhancing the adaptive capabilities of large language models is a critical pursuit in both research and application. Traditional fine-tuning methods require substantial data, computational resources, and specific capabilities, while in-context learning is limited by the need for appropriate demonstrations and efficient token usage. Inspired by the expression of in-context learned capabilities through task vectors and the concept of modular capability or knowledge, we propose ELICIT, a framework consisting of two modules designed to effectively store and reuse task vectors to enhance the diverse adaptive capabilities of models without additional training or inference tokens. Our comprehensive experiments and analysis demonstrate that our pipeline is highly transferable across different input formats, tasks, and model architectures. Externally storing and reusing vectors that represent in-context learned capabilities not only shows the potential to extract modular capabilities but also significantly enhances the performance, versatility, adaptability, and scalability of large language models, paving the way for more efficient and effective use of these models in a wide range of applications.