Saved in:
Bibliographic Details
Main Authors: Somu, Pranav, Balakrishnan, Advay, Kravtsov, Stepan, McDaniel, Aaron, Zutty, Jason
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.15649
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911687824637952
author Somu, Pranav
Balakrishnan, Advay
Kravtsov, Stepan
McDaniel, Aaron
Zutty, Jason
author_facet Somu, Pranav
Balakrishnan, Advay
Kravtsov, Stepan
McDaniel, Aaron
Zutty, Jason
contents Developing effective surrogates (performance predictors) for Neural Architecture Search (NAS) typically requires expensive fine-tuning or the engineering of complex representations. We propose a low-cost embedding strategy that leverages the inductive bias of Language Models (LMs) to eliminate these overheads. By representing architectures as PyTorch class definition text, we demonstrate that off-the-shelf LMs act as competitive feature extractors without NAS-specialized fine-tuning. The final predictor is constructed by passing the extracted Code-Oriented LM Embeddings (COLE) through a lightweight regression head. We also investigate strategies to improve embedding quality and utilization. Our experiments on the NAS-Bench-201 and einspace search spaces reveal that raw code inputs yield higher predictive performance than other text-based encodings (e.g., ONNX-to-text encodings) when using frozen LMs. We also observe COLE drives superior surrogate-assisted search using the BANANAS algorithm in NAS-Bench-201. When optimizing for CIFAR-100 performance, replacing structural path encodings with COLE for architecture representation allows for a 34% decrease in the evaluation budget required to reach within 1% of the fittest architecture in the search space (by test accuracy). As any neural architecture can be represented as code, these findings establish COLE as a versatile and efficient foundation for advancing NAS.
format Preprint
id arxiv_https___arxiv_org_abs_2605_15649
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Towards Code-Oriented LM Embeddings for Surrogate-Assisted Neural Architecture Search
Somu, Pranav
Balakrishnan, Advay
Kravtsov, Stepan
McDaniel, Aaron
Zutty, Jason
Machine Learning
Neural and Evolutionary Computing
Developing effective surrogates (performance predictors) for Neural Architecture Search (NAS) typically requires expensive fine-tuning or the engineering of complex representations. We propose a low-cost embedding strategy that leverages the inductive bias of Language Models (LMs) to eliminate these overheads. By representing architectures as PyTorch class definition text, we demonstrate that off-the-shelf LMs act as competitive feature extractors without NAS-specialized fine-tuning. The final predictor is constructed by passing the extracted Code-Oriented LM Embeddings (COLE) through a lightweight regression head. We also investigate strategies to improve embedding quality and utilization. Our experiments on the NAS-Bench-201 and einspace search spaces reveal that raw code inputs yield higher predictive performance than other text-based encodings (e.g., ONNX-to-text encodings) when using frozen LMs. We also observe COLE drives superior surrogate-assisted search using the BANANAS algorithm in NAS-Bench-201. When optimizing for CIFAR-100 performance, replacing structural path encodings with COLE for architecture representation allows for a 34% decrease in the evaluation budget required to reach within 1% of the fittest architecture in the search space (by test accuracy). As any neural architecture can be represented as code, these findings establish COLE as a versatile and efficient foundation for advancing NAS.
title Towards Code-Oriented LM Embeddings for Surrogate-Assisted Neural Architecture Search
topic Machine Learning
Neural and Evolutionary Computing
url https://arxiv.org/abs/2605.15649