RETRO pytorch
0.3.9
Implementación de RETRO, la red de atención basada en recuperación de Deepmind, en Pytorch. Esto se desviará ligeramente del papel, utilizando incrustaciones rotativas para la codificación posicional relativa, así como la biblioteca Faiss en lugar de Scann.
Esta biblioteca aprovecha autofaiss para crear el índice y calcular los k vecinos más cercanos para todos los fragmentos.
Entrada de blog explicativa de Jay Alammar
El punto de venta de este enfoque de recuperación es alcanzar el rendimiento de GPT-3 con 10 veces menos parámetros. Definitivamente se merece más investigación en esta área.
También he incluido las características necesarias para escalar el transformador de recuperación a 1000 capas, si hay que creer en las afirmaciones del artículo de DeepNet.
Actualización: Alguien en Reddit me regaló un premio de oro. No estoy seguro de qué es, ¡pero gracias!
Actualización: Deepnorm ha sido validado a escala en un modelo 130B de Tsinghua. Ahora se recomienda entrenar con use_deepnet configurado en True
$ pip install retro-pytorch import torch
from retro_pytorch import RETRO
retro = RETRO (
chunk_size = 64 , # the chunk size that is indexed and retrieved (needed for proper relative positions as well as causal chunked cross attention)
max_seq_len = 2048 , # max sequence length
enc_dim = 896 , # encoder model dim
enc_depth = 2 , # encoder depth
dec_dim = 796 , # decoder model dim
dec_depth = 12 , # decoder depth
dec_cross_attn_layers = ( 3 , 6 , 9 , 12 ), # decoder cross attention layers (with causal chunk cross attention)
heads = 8 , # attention heads
dim_head = 64 , # dimension per head
dec_attn_dropout = 0.25 , # decoder attention dropout
dec_ff_dropout = 0.25 , # decoder feedforward dropout
use_deepnet = True # turn on post-normalization with DeepNet residual scaling and initialization, for scaling to 1000 layers
)
seq = torch . randint ( 0 , 20000 , ( 2 , 2048 + 1 )) # plus one since it is split into input and labels for training
retrieved = torch . randint ( 0 , 20000 , ( 2 , 32 , 2 , 128 )) # retrieved tokens - (batch, num chunks, num retrieved neighbors, retrieved chunk with continuation)
loss = retro ( seq , retrieved , return_loss = True )
loss . backward ()
# do above for many steps El objetivo de TrainingWrapper es procesar una carpeta de documentos de texto en las matrices numpy memmapped necesarias para comenzar a entrenar RETRO .
import torch
from retro_pytorch import RETRO , TrainingWrapper
# instantiate RETRO, fit it into the TrainingWrapper with correct settings
retro = RETRO (
max_seq_len = 2048 , # max sequence length
enc_dim = 896 , # encoder model dimension
enc_depth = 3 , # encoder depth
dec_dim = 768 , # decoder model dimensions
dec_depth = 12 , # decoder depth
dec_cross_attn_layers = ( 1 , 3 , 6 , 9 ), # decoder cross attention layers (with causal chunk cross attention)
heads = 8 , # attention heads
dim_head = 64 , # dimension per head
dec_attn_dropout = 0.25 , # decoder attention dropout
dec_ff_dropout = 0.25 # decoder feedforward dropout
). cuda ()
wrapper = TrainingWrapper (
retro = retro , # path to retro instance
knn = 2 , # knn (2 in paper was sufficient)
chunk_size = 64 , # chunk size (64 in paper)
documents_path = './text_folder' , # path to folder of text
glob = '**/*.txt' , # text glob
chunks_memmap_path = './train.chunks.dat' , # path to chunks
seqs_memmap_path = './train.seq.dat' , # path to sequence data
doc_ids_memmap_path = './train.doc_ids.dat' , # path to document ids per chunk (used for filtering neighbors belonging to same document)
max_chunks = 1_000_000 , # maximum cap to chunks
max_seqs = 100_000 , # maximum seqs
knn_extra_neighbors = 100 , # num extra neighbors to fetch
max_index_memory_usage = '100m' ,
current_memory_available = '1G'
)
# get the dataloader and optimizer (AdamW with all the correct settings)
train_dl = iter ( wrapper . get_dataloader ( batch_size = 2 , shuffle = True ))
optim = wrapper . get_optimizer ( lr = 3e-4 , wd = 0.01 )
# now do your training
# ex. one gradient step
seq , retrieved = map ( lambda t : t . cuda (), next ( train_dl ))
# seq - (2, 2049) - 1 extra token since split by seq[:, :-1], seq[:, 1:]
# retrieved - (2, 32, 2, 128) - 128 since chunk + continuation, each 64 tokens
loss = retro (
seq ,
retrieved ,
return_loss = True
)
# one gradient step
loss . backward ()
optim . step ()
optim . zero_grad ()
# do above for many steps, then ...
# topk sampling with retrieval at chunk boundaries
sampled = wrapper . generate ( filter_thres = 0.9 , temperature = 1.0 ) # (1, <2049) terminates early if all <eos>
# or you can generate with a prompt, knn retrieval for initial chunks all taken care of
prompt = torch . randint ( 0 , 1000 , ( 1 , 128 )) # start with two chunks worth of sequence
sampled = wrapper . generate ( prompt , filter_thres = 0.9 , temperature = 1.0 ) # (1, <2049) terminates early if all <eos> Si desea forzar un reprocesamiento de los datos de entrenamiento, simplemente ejecute su script con un indicador de entorno REPROCESS=1 como tal.
$ REPROCESS=1 python train.py La clase RETRODataset acepta rutas a una serie de matrices numpy con memoria que contienen los fragmentos, el índice del primer fragmento de la secuencia que se va a entrenar (en el decodificador RETRO) y los índices precalculados de los k vecinos más cercanos por fragmento.
Puede utilizar esto para reunir fácilmente los datos para el entrenamiento RETRO , si no desea utilizar el TrainingWrapper de arriba.
Además, todas las funciones necesarias para crear los datos memmappeados necesarios se encuentran en las secciones siguientes.
import torch
from torch . utils . data import DataLoader
from retro_pytorch import RETRO , RETRODataset
# mock data constants
import numpy as np
NUM_CHUNKS = 1000
CHUNK_SIZE = 64
NUM_SEQS = 100
NUM_NEIGHBORS = 2
def save_memmap ( path , tensor ):
f = np . memmap ( path , dtype = tensor . dtype , mode = 'w+' , shape = tensor . shape )
f [:] = tensor
del f
# generate mock chunk data
save_memmap (
'./train.chunks.dat' ,
np . int32 ( np . random . randint ( 0 , 8192 , size = ( NUM_CHUNKS , CHUNK_SIZE + 1 )))
)
# generate nearest neighbors for each chunk
save_memmap (
'./train.chunks.knn.dat' ,
np . int32 ( np . random . randint ( 0 , 1000 , size = ( NUM_CHUNKS , NUM_NEIGHBORS )))
)
# generate seq data
save_memmap (
'./train.seq.dat' ,
np . int32 ( np . random . randint ( 0 , 128 , size = ( NUM_SEQS ,)))
)
# instantiate dataset class
# which constructs the sequence and neighbors from memmapped chunk and neighbor information
train_ds = RETRODataset (
num_sequences = NUM_SEQS ,
num_chunks = NUM_CHUNKS ,
num_neighbors = NUM_NEIGHBORS ,
chunk_size = CHUNK_SIZE ,
seq_len = 2048 ,
chunk_memmap_path = './train.chunks.dat' ,
chunk_nn_memmap_path = './train.chunks.knn.dat' ,
seq_memmap_path = './train.seq.dat'
)
train_dl = iter ( DataLoader ( train_ds , batch_size = 2 ))
# one forwards and backwards
retro = RETRO (
max_seq_len = 2048 , # max sequence length
enc_dim = 896 , # encoder model dimension
enc_depth = 3 , # encoder depth
dec_dim = 768 , # decoder model dimensions
dec_depth = 12 , # decoder depth
dec_cross_attn_layers = ( 1 , 3 , 6 , 9 ), # decoder cross attention layers (with causal chunk cross attention)
heads = 8 , # attention heads
dim_head = 64 , # dimension per head
dec_attn_dropout = 0.25 , # decoder attention dropout
dec_ff_dropout = 0.25 # decoder feedforward dropout
). cuda ()
seq , retrieved = map ( lambda t : t . cuda (), next ( train_dl ))
# seq - (2, 2049) - 1 extra token since split by seq[:, :-1], seq[:, 1:]
# retrieved - (2, 32, 2, 128) - 128 since chunk + continuation, each 64 tokens
loss = retro (
seq ,
retrieved ,
return_loss = True
)
loss . backward ()Este repositorio utilizará el tokenizador predeterminado (frase) para la versión en mayúsculas de BERT. Las incrustaciones se obtendrán del BERT básico y pueden ser una representación agrupada media enmascarada o el token CLS.
ex. representación agrupada media enmascarada
from retro_pytorch . retrieval import bert_embed , tokenize
ids = tokenize ([
'hello world' ,
'foo bar'
])
embeds = bert_embed ( ids ) # (2, 768) - 768 is hidden dimension of BERTex. Representación del token CLS
from retro_pytorch . retrieval import bert_embed , tokenize
ids = tokenize ([
'hello world' ,
'foo bar'
])
embeds = bert_embed ( ids , return_cls_repr = True ) # (2, 768) Cree sus fragmentos e índices de inicio de fragmentos (para calcular rangos de secuencia para entrenamiento autorregresivo) usando text_folder_to_chunks_
from retro_pytorch . retrieval import text_folder_to_chunks_
stats = text_folder_to_chunks_ (
folder = './text_folder' ,
glob = '**/*.txt' ,
chunks_memmap_path = './train.chunks.dat' ,
seqs_memmap_path = './train.seq.dat' ,
doc_ids_memmap_path = './train.doc_ids.dat' , # document ids are needed for filtering out neighbors belonging to same document appropriately during computation of nearest neighbors
chunk_size = 64 ,
seq_len = 2048 ,
max_chunks = 1_000_000 ,
max_seqs = 100_000
)
# {'chunks': <number of chunks>, 'docs': <number of documents>, 'seqs': <number of sequences>} Puede convertir su matriz numpy de fragmentos memmapped en incrustaciones y un índice faiss con un solo comando
from retro_pytorch . retrieval import chunks_to_index_and_embed
index , embeddings = chunks_to_index_and_embed (
num_chunks = 1000 ,
chunk_size = 64 ,
chunk_memmap_path = './train.chunks.dat'
)
query_vector = embeddings [: 1 ] # use first embedding as query
_ , indices = index . search ( query_vector , k = 2 ) # fetch 2 neighbors, first indices should be self
neighbor_embeddings = embeddings [ indices ] # (1, 2, 768) También puede calcular directamente el archivo vecino más cercano necesario para el entrenamiento, con el comando chunks_to_precalculated_knn_
from retro_pytorch . retrieval import chunks_to_precalculated_knn_
chunks_to_precalculated_knn_ (
num_chunks = 1000 ,
chunk_size = 64 ,
chunk_memmap_path = './train.chunks.dat' , # path to main chunks dataset
doc_ids_memmap_path = './train.doc_ids.dat' , # path to document ids created by text_folder_to_chunks_, used for filtering out neighbors that belong to the same document
num_nearest_neighbors = 2 , # number of nearest neighbors you'd like to use
num_extra_neighbors = 10 # fetch 10 extra neighbors, in the case that fetched neighbors are frequently from same document (filtered out)
)
# nearest neighbor info saved to ./train.chunks.knn.dat @misc { borgeaud2022improving ,
title = { Improving language models by retrieving from trillions of tokens } ,
author = { Sebastian Borgeaud and Arthur Mensch and Jordan Hoffmann and Trevor Cai and Eliza Rutherford and Katie Millican and George van den Driessche and Jean-Baptiste Lespiau and Bogdan Damoc and Aidan Clark and Diego de Las Casas and Aurelia Guy and Jacob Menick and Roman Ring and Tom Hennigan and Saffron Huang and Loren Maggiore and Chris Jones and Albin Cassirer and Andy Brock and Michela Paganini and Geoffrey Irving and Oriol Vinyals and Simon Osindero and Karen Simonyan and Jack W. Rae and Erich Elsen and Laurent Sifre } ,
year = { 2022 } ,
eprint = { 2112.04426 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CL }
} @misc { su2021roformer ,
title = { RoFormer: Enhanced Transformer with Rotary Position Embedding } ,
author = { Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu } ,
year = { 2021 } ,
eprint = { 2104.09864 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CL }
} @article { Wang2022DeepNetST ,
title = { DeepNet: Scaling Transformers to 1, 000 Layers } ,
author = { Hongyu Wang and Shuming Ma and Li Dong and Shaohan Huang and Dongdong Zhang and Furu Wei } ,
journal = { ArXiv } ,
year = { 2022 } ,
volume = { abs/2203.00555 }
} @misc { zhang2021sparse ,
title = { Sparse Attention with Linear Units } ,
author = { Biao Zhang and Ivan Titov and Rico Sennrich } ,
year = { 2021 } ,
eprint = { 2104.07012 } ,
archivePrefix = { arXiv } ,
primaryClass = { cs.CL }
}Siempre considero la vida adulta como la recuperación continua de la infancia. -Umberto Eco
