% Generated by roxygen2: do not edit by hand

% Please edit documentation in R/gptneox.R

\name{gptneox}

\alias{gptneox}

\alias{gptneox_from_config}

\alias{gptneox_from_pretrained}

\title{GPT NeoX}

\usage{

gptneox(

vocab_size = 50432,

n_embd = 6144,

n_head = 64,

n_layer = 44,

max_pos = 2048,

n_rot = 0.25

)

gptneox_from_config(identifier, revision = "main")

gptneox_from_pretrained(identifier, revision = "main")

}

\arguments{

\item{vocab_size}{An integer indicating the size of the vocabulary or the number

of unique tokens in the input data.}

\item{n_embd}{An integer specifying the dimensionality of the embedding vectors.}

\item{n_head}{An integer representing the number of attention heads in the

multi-head attention mechanism.}

\item{n_layer}{An integer indicating the number of layers in the deep learning model.}

\item{max_pos}{An integer specifying the maximum position encoding value or

the maximum sequence length.}

\item{n_rot}{An integer indicating the number dimensions used in the rotary

position embedding. Can also be a float \verb{0 < n_rot < 1} indicating the fraction

of \code{n_embd}.}

\item{identifier}{A string representing the identifier or name of the pre-trained

model in the Hugging Face model hub.}

\item{revision}{A string specifying the revision or version of the pre-trained

model in the Hugging Face model hub.}

}

\value{

An initialized \code{\link[torch:nn_module]{torch::nn_module()}}.

}

\description{

Initializes a GPTNeoX like model

}

\section{Functions}{

\itemize{

\item \code{gptneox_from_config()}: Initializes a gptneox model using a configuration defined in HF Hub

\item \code{gptneox_from_pretrained()}: Initializes the gptneox model and load pre-trained weights from HF hub.

}}