The Mamba-type neural networks have gained significant popularity recently. To effectively and efficiently establish model architectures of Mamba, it is natural to introduce Neural Architecture Search (NAS) methods into Mamba. However, existing NAS methods tailored for Mamba are training-based, leading to substantial time and computational resource expenditure. To address this issue, and considering that Mamba2 is an improved version of the original Mamba, we propose a trainingfree NAS method specifically designed for Mamba2. Based on rank collapse in stacked State Space Duality (SSD) blocks, we design a proxy that only requires the computation of the transformation matrix and its gradient between two tensors within the network. Additionally, we develop a corresponding search space and introduce a novel approach for determining adjustable hyperparameter ranges. Experimental results show that our method outperforms all existing training-free NAS approaches in terms of both ranking correlation and the performance of search results for Mamba2 architecture. To the best of our knowledge, this is the first training-free NAS method designed for Mamba-type architectures.

Inourexperiments, disentanglement models andemergentlanguage (EL) models use the same architectures for the convolutional encoder and decoder. On the dSprites dataset, we use 500 samples to train linear or GBT readout models for classification and regressions tasks. Their generalization performance is given in Table 3. The attribute values are expected to generalize perfectly with linear orGBT readout models. Compositional latent variables may not be the best representations for downstream tasks.

A.1 NASSearchSpaces NASBench-1011 introduces a large and expressive search space with 423k unique convolutional neural architectures and training statistics on CIFAR-10. NASBench-2012 contains the training statistics of15,625 architectures across three different datasets, including CIFAR-10, CIFAR-100, and Tiny-ImageNet-16. NASBench-NLP5 [10] is an NLP neural architecture search space, including 14k recurrent cells trained on the Penn Treebank (PTB)[11]dataset. Thegenerated noise maps aredirectly multiplied bythelevelwhich canbeselected from 0 to 1 with a step of 0.1. Other settings are already described in Section 3.1.2.

Most existing neural architecture search (NAS) algorithms are dedicated to and evaluated bythedownstream tasks, e.g.,image classification incomputer vision.

The ResNeXt-A and ResNeXt-B have different channel-number and group-convolution settings. Initialize an empty population queue, Q_pop // The maximum population is P Initialize an empty set, history // Will contain all visited individuals for i = 1, 2,, P do new _individual RandomInit() new _individual.fitness Eval(new _individual) Enqueue(Q _pop, new _individual) // Add individual to the right of Q _pop Add new _ individual to history end // Evolve for T _iter for i = 1, 2,, T_iter do Initialize an empty set, sample _ set for i = 1, 2,, S do Add an individual to sample _ set from Q_ pop without replacement. A.2.2 Proxy T ask Search Figure 1: The configuration of a task in JSON style and the illustration of task mutation. The configuration of a task is shown in Figure 1.

Extensive experiments across 13 CNN search spaces and one NLP space demonstrate the remarkable efficiency of GenNAS using regression, in terms of both evaluating the neural architectures (quantified by the ranking correlation Spearman's

Neural architecture search (NAS) enables the automatic design of neural network models. However, training the candidates generated by the search algorithm for performance evaluation incurs considerable computational overhead. Our method, dubbed nasgraph, remarkably reduces the computational costs by converting neural architectures to graphs and using the average degree, a graph measure, as the proxy in lieu of the evaluation metric. Our training-free NAS method is data-agnostic and light-weight. It can find the best architecture among 200 randomly sampled architectures from NAS-Bench201 in 217 CPU seconds. Besides, our method is able to achieve competitive performance on various datasets including NASBench-101, NASBench-201, and NDS search spaces. We also demonstrate that nasgraph generalizes to more challenging tasks on Micro TransNAS-Bench-101.

Neural Architecture Search (NAS) has become a de facto approach in the recent trend of AutoML to design deep neural networks (DNNs). Efficient or near-zero-cost NAS proxies are further proposed to address the demanding computational issues of NAS, where each candidate architecture network only requires one iteration of backpropagation. The values obtained from the proxies are considered the predictions of architecture performance on downstream tasks. However, two significant drawbacks hinder the extended usage of Efficient NAS proxies. (1) Efficient proxies are not adaptive to various search spaces. (2) Efficient proxies are not extensible to multi-modality downstream tasks. Based on the observations, we design a Extensible proxy (Eproxy) that utilizes self-supervised, few-shot training (i.e., 10 iterations of backpropagation) which yields near-zero costs. The key component that makes Eproxy efficient is an untrainable convolution layer termed barrier layer that add the non-linearities to the optimization spaces so that the Eproxy can discriminate the performance of architectures in the early stage. Furthermore, to make Eproxy adaptive to different downstream tasks/search spaces, we propose a Discrete Proxy Search (DPS) to find the optimized training settings for Eproxy with only handful of benchmarked architectures on the target tasks. Our extensive experiments confirm the effectiveness of both Eproxy and Eproxy+DPS. Code is available at https://github.com/leeyeehoo/GenNAS-Zero.

Realistic use of neural networks often requires adhering to multiple constraints on latency, energy and memory among others. A popular approach to find fitting networks is through constrained Neural Architecture Search (NAS). However, previous methods use complicated predictors for the accuracy of the network. Those predictors are hard to interpret and sensitive to many hyperparameters to be tuned, hence, the resulting accuracy of the generated models is often harmed. In this work we resolve this by introducing Interpretable Integer Quadratic programming Neural Architecture Search (IQNAS), that is based on an accurate and simple quadratic formulation of both the accuracy predictor and the expected resource requirement, together with a scalable search method with theoretical guarantees. The simplicity of our proposed predictor together with the intuitive way it is constructed bring interpretability through many insights about the contribution of different design choices. For example, we find that in the examined search space, adding depth and width is more effective at deeper stages of the network and at the beginning of each resolution stage. Our experiments show that IQNAS generates comparable to or better architectures than other state-of-the-art NAS methods within a reduced search cost for each additional generated network, while strictly satisfying the resource constraints.

Most existing neural architecture search (NAS) algorithms are dedicated to the downstream tasks, e.g., image classification in computer vision. However, extensive experiments have shown that, prominent neural architectures, such as ResNet in computer vision and LSTM in natural language processing, are generally good at extracting patterns from the input data and perform well on different downstream tasks. These observations inspire us to ask: Is it necessary to use the performance of specific downstream tasks to evaluate and search for good neural architectures? Can we perform NAS effectively and efficiently while being agnostic to the downstream task? In this work, we attempt to affirmatively answer the above two questions and improve the state-of-the-art NAS solution by proposing a novel and generic NAS framework, termed Generic NAS (GenNAS). GenNAS does not use task-specific labels but instead adopts \textit{regression} on a set of manually designed synthetic signal bases for architecture evaluation. Such a self-supervised regression task can effectively evaluate the intrinsic power of an architecture to capture and transform the input signal patterns, and allow more sufficient usage of training samples. We then propose an automatic task search to optimize the combination of synthetic signals using limited downstream-task-specific labels, further improving the performance of GenNAS. We also thoroughly evaluate GenNAS's generality and end-to-end NAS performance on all search spaces, which outperforms almost all existing works with significant speedup.