基础模型适配

In this work, we analyze both theoretically and empirically the effect of tied input-output embeddings—a popular technique that reduces the model size while often improving training. Interestingly, we found that this technique is connected to Harris (1954)’s distributional hypothesis—often portrayed by the famous Firth (1957)’s quote “a word is characterized by the company it keeps”. Specifically, our findings indicate that words (or, more broadly, symbols) with similar semantics tend to be encoded in similar input embeddings, while words that appear in similar contexts are encoded in similar output embeddings (thus explaining the semantic space arising in input and output embedding of foundational language models). As a consequence of these findings, the tying of the input and output embeddings is encouraged only when the distributional hypothesis holds for the underlying data. These results also provide insight into the embeddings of foundation language models (which are known to be semantically organized). Further, we complement the theoretical findings with several experiments supporting the claims.

Parameter-efficient finetuning (PEFT) has become ubiquitous to adapt foundation models to downstream task requirements while retaining their generalization ability. However, the amount of additionally introduced parameters and compute for successful adaptation and hyperparameter searches can explode quickly, especially when deployed at scale to serve numerous individual requests. To ensure effective, parameter-efficient, and hyperparameter-robust adaptation, we propose the *ETHER* transformation family, which performs Efficient fineTuning via HypErplane Reflections. By design, *ETHER* transformations require *a minimal number of parameters*, are *less likely to deteriorate model performance*, and exhibit *robustness to hyperparameter and learning rate choices*. In particular, we introduce *ETHER* and its relaxation *ETHER+*, which match or outperform existing PEFT methods with significantly fewer parameters ($\sim$$10$-$100$ times lower than LoRA or OFT) across multiple image synthesis and natural language tasks without *exhaustive hyperparameter tuning*. Finally, we investigate the recent emphasis on Hyperspherical Energy retention for adaptation and raise questions on its practical utility. The code is available at https://github.com/mwbini/ether.

The objective of traffic prediction is to accurately forecast and analyze the dynamics of transportation patterns, considering both space and time. However, the presence of distribution shift poses a significant challenge in this field, as existing models struggle to generalize well when faced with test data that significantly differs from the training distribution. To tackle this issue, this paper introduces a simple and universal spatio-temporal prompt-tuning framework-FlashST, which adapts pre-trained models to the specific characteristics of diverse downstream datasets, improving generalization in diverse traffic prediction scenarios. Specifically, the FlashST framework employs a lightweight spatio-temporal prompt network for in-context learning, capturing spatio-temporal invariant knowledge and facilitating effective adaptation to diverse scenarios. Additionally, we incorporate a distribution mapping mechanism to align the data distributions of pre-training and downstream data, facilitating effective knowledge transfer in spatio-temporal forecasting. Empirical evaluations demonstrate the effectiveness of our FlashST across different spatio-temporal prediction tasks using diverse urban datasets. Code is available at [https://github.com/HKUDS/FlashST](https://github.com/HKUDS/FlashST).

We investigate the computational limits of the memory retrieval dynamics of modern Hopfield models from the fine-grained complexity analysis. Our key contribution is the characterization of a phase transition behavior in the efficiency of all possible modern Hopfield models based on the norm of patterns. Specifically, we establish an upper bound criterion for the norm of input query patterns and memory patterns. Only below this criterion, sub-quadratic (efficient) variants of the modern Hopfield model exist, assuming the Strong Exponential Time Hypothesis (SETH). To showcase our theory, we provide a formal example of efficient constructions of modern Hopfield models using low-rank approximation when the efficient criterion holds. This includes a derivation of a lower bound on the computational time, scaling linearly with $\max$\{ \# of stored memory patterns, length of input query sequence\}. In addition, we prove its memory retrieval error bound and exponential memory capacity.

While overparameterization in machine learning models offers great benefits in terms of optimization and generalization, it also leads to increased computational requirements as model sizes grow. In this work, we show that by leveraging the inherent low-dimensional structures of data and compressible dynamics within the model parameters, we can reap the benefits of overparameterization without the computational burdens. In practice, we demonstrate the effectiveness of this approach for deep low-rank matrix completion as well as fine-tuning language models. Our approach is grounded in theoretical findings for deep overparameterized low-rank matrix recovery, where we show that the learning dynamics of each weight matrix are confined to an invariant low-dimensional subspace. Consequently, we can construct and train compact, highly compressed factorizations possessing the same benefits as their overparameterized counterparts. In the context of deep matrix completion, our technique substantially improves training efficiency while retaining the advantages of overparameterization. For language model fine-tuning, we propose a method called "Deep LoRA", which improves the existing low-rank adaptation (LoRA) technique, leading to reduced overfitting and a simplified hyperparameter setup, while maintaining comparable efficiency. We validate the effectiveness of Deep LoRA on natural language tasks, particularly when fine-tuning with limited data.

Parameter-efficient fine-tuning optimizes large, pre-trained foundation models by updating a subset of parameters; in this class, Low-Rank Adaptation (LoRA) is particularly effective. Inspired by an effort to investigate the different roles of LoRA matrices during fine-tuning, this paper characterizes and leverages unexpected asymmetry in the importance of low-rank adapter matrices. Specifically, when updating the parameter matrices of a neural network by adding a product $BA$, we observe that the $B$ and $A$ matrices have distinct functions: $A$ extracts features from the input, while $B$ uses these features to create the desired output. Based on this observation, we demonstrate that fine-tuning $B$ is inherently more effective than fine-tuning $A$, and that a random untrained $A$ should perform nearly as well as a fine-tuned one. Using an information-theoretic lens, we also bound the generalization of low-rank adapters, showing that the parameter savings of exclusively training $B$ improves the bound. We support our conclusions with experiments on RoBERTa, BART-Large, LLaMA-2, and ViTs. The code and data is available at https://github.com/Jiacheng-Zhu-AIML/AsymmetryLoRA

As a more challenging domain adaptation setting, universal domain adaptation (UniDA) introduces category shift on top of domain shift, which needs to identify unknown category in the target domain and avoid misclassifying target samples into source private categories. To this end, we propose a novel UniDA approach named Batch Singular value Polarization and Weighted Semantic Augmentation (BSP-WSA). Specifically, we adopt an adversarial classifier to identify the target unknown category and align feature distributions between the two domains. Then, we propose to perform SVD on the classifier's outputs to maximize larger singular values while minimizing those smaller ones, which could prevent target samples from being wrongly assigned to source private classes. To better bridge the domain gap, we propose a weighted semantic augmentation approach for UniDA to generate data on common categories between the two domains. Extensive experiments on three benchmarks demonstrate that BSP-WSA could outperform existing state-of-the-art UniDA approaches.

Low-rank adaptation (LoRA) has become the standard approach for parameter-efficient fine-tuning of large language models (LLM), but our theoretical understanding of LoRA has been limited. In this work, we theoretically analyze LoRA fine-tuning in the neural tangent kernel (NTK) regime with $N$ data points, showing: (i) full fine-tuning (without LoRA) admits a low-rank solution of rank $r\lesssim \sqrt{N}$; (ii) using LoRA with rank $r\gtrsim \sqrt{N}$ eliminates spurious local minima, allowing gradient descent to find the low-rank solutions; (iii) the low-rank solution found using LoRA generalizes well.

Acquiring a multi-task imitation policy in 3D manipulation poses challenges in terms of scene understanding and action prediction. Current methods employ both 3D representation and multi-view 2D representation to predict the poses of the robot’s end-effector. However, they still require a considerable amount of high-quality robot trajectories, and suffer from limited generalization in unseen tasks and inefficient execution in long-horizon reasoning. In this paper, we propose **SAM-E**, a novel architecture for robot manipulation by leveraging a vision-foundation model for generalizable scene understanding and sequence imitation for long-term action reasoning. Specifically, we adopt Segment Anything (SAM) pre-trained on a huge number of images and promptable masks as the foundation model for extracting task-relevant features, and employ parameter-efficient fine-tuning on robot data for a better understanding of embodied scenarios. To address long-horizon reasoning, we develop a novel multi-channel heatmap that enables the prediction of the action sequence in a single pass, notably enhancing execution efficiency. Experimental results from various instruction-following tasks demonstrate that SAM-E achieves superior performance with higher execution efficiency compared to the baselines, and also significantly improves generalization in few-shot adaptation to new tasks.

Vision-language models (VLMs) have emerged as formidable tools, showing their strong capability in handling various open-vocabulary tasks in image recognition, text-driven visual content generation, and visual chatbots, to name a few. In recent years, considerable efforts and resources have been devoted to adaptation methods for improving downstream performance of VLMs, particularly on parameter-efficient fine-tuning methods like prompt learning. However, a crucial aspect that has been largely overlooked is the confidence calibration problem in fine-tuned VLMs, which could greatly reduce reliability when deploying such models in the real world. This paper bridges the gap by systematically investigating the confidence calibration problem in the context of prompt learning and reveals that existing calibration methods are insufficient to address the problem, especially in the open-vocabulary setting. To solve the problem, we present a simple and effective approach called Distance-Aware Calibration (DAC), which is based on scaling the temperature using as guidance the distance between predicted text labels and base classes. The experiments with 7 distinct prompt learning methods applied across 11 diverse downstream datasets demonstrate the effectiveness of DAC, which achieves high efficacy without sacrificing the inference speed.

In many applications, the labeled data at the learner's disposal is subject to privacy constraints and is relatively limited. To derive a more accurate predictor for the target domain, it is often beneficial to leverage publicly available labeled data from an alternative domain, somewhat close to the target domain. This is the modern problem of supervised domain adaptation from a public source to a private target domain. We present two $(\epsilon, \delta)$-differentially private adaptation algorithms for supervised adaptation, for which we make use of a general optimization problem, recently shown to benefit from favorable theoretical learning guarantees. Our first algorithm is designed for regression with linear predictors and shown to solve a convex optimization problem. Our second algorithm is a more general solution for loss functions that may be non-convex but Lipschitz and smooth. While our main objective is a theoretical analysis, we also report the results of several experiments. We first show that the non-private versions of our algorithms match state-of-the-art performance in supervised adaptation and that for larger values of the target sample size or $\epsilon$, the performance of our private algorithms remains close to that of their non-private counterparts.

Training large transformers is slow, but recent innovations on GPU architecture give us an advantage. NVIDIA Ampere GPUs can execute a fine-grained 2:4 sparse matrix multiplication twice as fast as its dense equivalent. In the light of this property, we comprehensively investigate the feasibility of accelerating feed-forward networks (FFNs) of transformers in pre-training. First, we define a ``flip rate'' to monitor the stability of a 2:4 training process. Utilizing this metric, we propose three techniques to preserve accuracy: to modify the sparse-refined straight-through estimator by applying the masked decay term on gradients, to determine a feasible decay factor in warm-up stage, and to enhance the model's quality by a dense fine-tuning procedure near the end of pre-training. Besides, we devise two techniques to practically accelerate training: to calculate transposable 2:4 masks by convolution, and to accelerate gated activation functions by reducing GPU L2 cache miss. Experiments show that our 2:4 sparse training algorithm achieves similar convergence to dense training algorithms on several transformer pre-training tasks, while actual acceleration can be observed on different shapes of transformer block apparently. Our toolkit is available at https://github.com/huyz2023/2by4-pretrain.

Decision making demands intricate interplay between perception, memory, and reasoning to discern optimal policies. Conventional approaches to decision making face challenges related to low sample efficiency and poor generalization. In contrast, foundation models in language and vision have showcased rapid adaptation to diverse new tasks. Therefore, we advocate for the construction of foundation agents as a transformative shift in the learning paradigm of agents. This proposal is underpinned by the formulation of foundation agents with their fundamental characteristics and challenges motivated by the success of large language models (LLMs). Moreover, we specify the roadmap of foundation agents from large interactive data collection or generation, to self-supervised pretraining and adaptation, and knowledge and value alignment with LLMs. Lastly, we pinpoint critical research questions derived from the formulation and delineate trends for foundation agents supported by real-world use cases, addressing both technical and theoretical aspects to propel the field towards a more comprehensive and impactful future.

Motivated by recent advances in large language models for Natural Language Processing (NLP), we design a time-series foundation model for forecasting whose out-of-the-box zero-shot performance on a variety of public datasets comes close to the accuracy of state-of-the-art supervised forecasting models for each individual dataset. Our model is based on pretraining a decoder style attention model with input patching, using a large time-series corpus comprising both real-world and synthetic datasets. Experiments on a diverse set of previously unseen forecasting datasets suggests that the model can yield accurate zero-shot forecasts across different domains, forecasting horizons and temporal granularities.

With the emergence of pretrained vision-language models (VLMs), considerable efforts have been devoted to fine-tuning them for downstream tasks. Despite the progress made in designing efficient fine-tuning methods, such methods require access to the model's parameters, which can be challenging as model owners often opt to provide their models as a black box to safeguard model ownership. This paper proposes a **C**ollabo**ra**tive **F**ine-**T**uning (**CraFT**) approach for fine-tuning black-box VLMs to downstream tasks, where one only has access to the input prompts and the output predictions of the model. CraFT comprises two modules, a prompt generation module for learning text prompts and a prediction refinement module for enhancing output predictions in residual style. Additionally, we introduce an auxiliary prediction-consistent loss to promote consistent optimization across these modules. These modules are optimized by a novel collaborative training algorithm. Extensive experiments on few-shot classification over 15 datasets demonstrate the superiority of CraFT. The results show that CraFT achieves a decent gain of about 12% with 16-shot datasets and only 8,000 queries. Moreover, CraFT trains faster and uses only about 1/80 of the memory footprint for deployment, while sacrificing only 1.62% compared to the white-box method. Our code is publicly available at https://github.com/mrflogs/CraFT.

Foundation models are powerful technologies: how they are released publicly directly shapes their societal impact. In this position paper, we focus on *open* foundation models, defined here as those with broadly available model weights (e.g., Llama 3, Stable Diffusion XL). We identify five distinctive properties (e.g., greater customizability, poor monitoring) that mediate their benefits and risks. Open foundation models present significant benefits, with some caveats, that span innovation, competition, the distribution of decision-making power, and transparency. To understand their risks of misuse, we design a risk assessment framework for analyzing their *marginal risk*. Across several misuse vectors (e.g., cyberattacks, bioweapons), we find that current research is insufficient to effectively characterize the marginal risk of open foundation models relative to pre-existing technologies. The framework helps explain why the marginal risk is low in some cases, clarifies disagreements about misuse risks by revealing that past work has focused on different subsets of the framework with different assumptions, and articulates a way forward for more constructive debate. Overall, our work helps support a more grounded assessment of the societal impact of open foundation models by outlining what research is needed to empirically validate their theoretical benefits and risks.

Progress in designing channel codes has been driven by human ingenuity and, fittingly, has been sporadic. Polar codes, developed on the foundation of Arikan’s polarization kernel, represent the latest breakthrough in coding theory and have emerged as the state-of-the-art error-correction code for short-to-medium block length regimes. In an effort to automate the invention of good channel codes, especially in this regime, we explore a novel, non-linear generalization of Polar codes, which we call DeepPolar codes. DeepPolar codes extend the conventional Polar coding framework by utilizing a larger kernel size and parameterizing these kernels and matched decoders through neural networks. Our results demonstrate that these data-driven codes effectively leverage the benefits of a larger kernel size, resulting in enhanced reliability when compared to both existing neural codes and conventional Polar codes.

We study the problem of differentially private (DP) fine-tuning of large pre-trained models — a recent privacy-preserving approach suitable for solving downstream tasks with sensitive data. Existing work has demonstrated that high accuracy is possible under strong privacy constraint, yet requires significant computational overhead or modifications to the network architecture. We propose differentially private bias-term fine-tuning (DP-BiTFiT), which matches the state-of-the-art accuracy for DP algorithms and the efficiency of the standard BiTFiT. DP-BiTFiT is model agnostic (not modifying the network architecture), parameter efficient (only training about 0.1% of the parameters), and computation efficient (almost removing the overhead caused by DP, in both the time and space complexity). On a wide range of tasks, DP-BiTFiT is 2 - 30X faster and uses 2 - 8X less memory than DP full fine-tuning, even faster than the standard full fine-tuning. This amazing efficiency enables us to conduct DP fine-tuning on language and vision tasks with long-sequence texts and high-resolution images, which were computationally difficult using existing methods.

Machine learning models fail catastrophically under distribution shift, but a surprisingly effective way to empirically improve robustness to some types of shift (*e.g.*, Imagenet-A/C) is to use stronger open-vocabulary classifiers derived from foundation models. In this work, we first note that for shifts governed by spurious correlations (features spuriously correlated with the label on the training data, but not on test), the zero-shot and few-shot performance of foundation models is no better than ERM models, and remains unchanged when pretrained data/model size is scaled. Secondly, even in these situations, foundation models are quite accurate at predicting the value of the spurious feature. In a simplified setup, we theoretically analyze both these findings. Specifically, we show that during contrastive pretraining, the simplicity bias of foundation models tends to result in the learning of features that mostly rely on the spurious attribute, compared to more robust features. We leverage these observations to propose Prompting for Robustness (PfR) which first uses foundation models to zero-shot predict the spurious attribute on labeled examples, and then learns a classifier with balanced performance across different groups of labels and spurious attribute. Across 5 vision and language tasks, we show that PfR's performance nearly equals that of an oracle algorithm (group DRO) that leverages human labeled spurious attributes.

Semi-supervised learning (SSL) has witnessed remarkable progress, resulting in the emergence of numerous method variations. However, practitioners often encounter challenges when attempting to deploy these methods due to their subpar performance. In this paper, we present a novel SSL approach named FineSSL that significantly addresses this limitation by adapting pre-trained foundation models. We identify the aggregated biases and cognitive deviation problems inherent in foundation models, and propose a simple yet effective solution by imposing balanced margin softmax and decoupled label smoothing. Through extensive experiments, we demonstrate that FineSSL sets a new state of the art for SSL on multiple benchmark datasets, reduces the training cost by over six times, and can seamlessly integrate various fine-tuning and modern SSL algorithms. The source code is available at https://github.com/Gank0078/FineSSL.