Saved in:
Bibliographic Details
Main Author: Gokden, Burc
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2410.16703
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929554092720128
author Gokden, Burc
author_facet Gokden, Burc
contents We present the Large Language Model from Power Law Decoder Representations (PLDR-LLM), a language model that leverages non-linear and linear transformations through Power Law Graph Attention mechanism to generate well-defined deductive and inductive outputs. We pretrain the PLDR-LLMs of varying layer sizes with a small batch size of 32 and $\sim$8B tokens from the RefinedWeb dataset, and show that they achieve competitive performance in zero-shot and few-shot settings compared to scaled dot-product LLMs of similar model size reported in the literature. We show that deductive outputs of PLDR-LLMs can be used to compare model characteristics or improve the performance by introducing the Directed Acyclic Graph (DAG) loss as a metric and regularizer. Our results indicate that the initial maximum learning rate and warm-up steps have a lasting impact on deductive outputs throughout the pretraining. We provide a detailed description of PLDR-LLM architecture, its implementation and the pretraining procedure.
format Preprint
id arxiv_https___arxiv_org_abs_2410_16703
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle PLDR-LLM: Large Language Model from Power Law Decoder Representations
Gokden, Burc
Computation and Language
Artificial Intelligence
We present the Large Language Model from Power Law Decoder Representations (PLDR-LLM), a language model that leverages non-linear and linear transformations through Power Law Graph Attention mechanism to generate well-defined deductive and inductive outputs. We pretrain the PLDR-LLMs of varying layer sizes with a small batch size of 32 and $\sim$8B tokens from the RefinedWeb dataset, and show that they achieve competitive performance in zero-shot and few-shot settings compared to scaled dot-product LLMs of similar model size reported in the literature. We show that deductive outputs of PLDR-LLMs can be used to compare model characteristics or improve the performance by introducing the Directed Acyclic Graph (DAG) loss as a metric and regularizer. Our results indicate that the initial maximum learning rate and warm-up steps have a lasting impact on deductive outputs throughout the pretraining. We provide a detailed description of PLDR-LLM architecture, its implementation and the pretraining procedure.
title PLDR-LLM: Large Language Model from Power Law Decoder Representations
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2410.16703