Synergy: End-to-end Concept Model

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Zheng, Keli, Xie, Zerong
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916848049586176
author Zheng, Keli
Xie, Zerong
author_facet Zheng, Keli
Xie, Zerong
contents In this paper, we present Synergy, a language model that bridges different levels of abstraction in an end-to-end fashion through a learned routing mechanism. Focusing on low-level linguistic abstraction, we trained our model as a byte-level language model. Our model spontaneously learns to tokenize bytes, producing fewer concept tokens than Byte-level Byte Pair Encoder (BBPE) tokenizers while keeping comparable performance. By comparing with Llama3, we observed an advantage of Synergy under the same model scale and training dataset size. Further studies show that the middle part (the higher abstraction part) of our model performs better when positional encodings are removed, suggesting the emergence of position-independent concepts. These findings demonstrate the feasibility of tokenizer-free architectures, paving the way for more robust and flexible pipelines.
format Preprint
id arxiv_https___arxiv_org_abs_2507_12769
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Synergy: End-to-end Concept Model
Zheng, Keli
Xie, Zerong
Computation and Language
Artificial Intelligence
I.2.7
In this paper, we present Synergy, a language model that bridges different levels of abstraction in an end-to-end fashion through a learned routing mechanism. Focusing on low-level linguistic abstraction, we trained our model as a byte-level language model. Our model spontaneously learns to tokenize bytes, producing fewer concept tokens than Byte-level Byte Pair Encoder (BBPE) tokenizers while keeping comparable performance. By comparing with Llama3, we observed an advantage of Synergy under the same model scale and training dataset size. Further studies show that the middle part (the higher abstraction part) of our model performs better when positional encodings are removed, suggesting the emergence of position-independent concepts. These findings demonstrate the feasibility of tokenizer-free architectures, paving the way for more robust and flexible pipelines.
title Synergy: End-to-end Concept Model
topic Computation and Language
Artificial Intelligence
I.2.7
url https://arxiv.org/abs/2507.12769