From Symbolic Tasks to Code Generation: Diversification Yields Better Task Performers

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Dylan, Wang, Justin, Charton, Francois
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916266758897664
author Zhang, Dylan
Wang, Justin
Charton, Francois
author_facet Zhang, Dylan
Wang, Justin
Charton, Francois
contents Instruction tuning -- tuning large language models on instruction-output pairs -- is a promising technique for making models better adapted to the real world. Yet, the key factors driving the model's capability to understand and follow instructions not seen during training remain under-explored. Our investigation begins with a series of synthetic experiments within the theoretical framework of a Turing-complete algorithm called Markov algorithm, which allows fine-grained control over the instruction-tuning data. Generalization and robustness with respect to the training distribution emerge once a diverse enough set of tasks is provided, even though very few examples are provided for each task. We extend these initial results to a real-world application scenario of code generation and find that a more diverse instruction set, extending beyond code-related tasks, improves the performance of code generation. Our observations suggest that a more diverse semantic space for instruction-tuning sets greatly improves the model's ability to follow instructions and perform tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2405_19787
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle From Symbolic Tasks to Code Generation: Diversification Yields Better Task Performers
Zhang, Dylan
Wang, Justin
Charton, Francois
Computation and Language
Artificial Intelligence
Machine Learning
Logic in Computer Science
Programming Languages
Instruction tuning -- tuning large language models on instruction-output pairs -- is a promising technique for making models better adapted to the real world. Yet, the key factors driving the model's capability to understand and follow instructions not seen during training remain under-explored. Our investigation begins with a series of synthetic experiments within the theoretical framework of a Turing-complete algorithm called Markov algorithm, which allows fine-grained control over the instruction-tuning data. Generalization and robustness with respect to the training distribution emerge once a diverse enough set of tasks is provided, even though very few examples are provided for each task. We extend these initial results to a real-world application scenario of code generation and find that a more diverse instruction set, extending beyond code-related tasks, improves the performance of code generation. Our observations suggest that a more diverse semantic space for instruction-tuning sets greatly improves the model's ability to follow instructions and perform tasks.
title From Symbolic Tasks to Code Generation: Diversification Yields Better Task Performers
topic Computation and Language
Artificial Intelligence
Machine Learning
Logic in Computer Science
Programming Languages
url https://arxiv.org/abs/2405.19787