CoLT5 attention
0.11.1
Implementación de la atención eficiente enrutada condicionalmente en la arquitectura CoLT5 propuesta, en Pytorch.
Utilizaron el descenso de coordenadas de este artículo (algoritmo principal originalmente de Wright et al) para enrutar un subconjunto de tokens para ramas "más pesadas" de los bloques de atención y avance.
Actualización: no estoy seguro de cómo se utilizan las puntuaciones normalizadas de enrutamiento para los valores-clave. Improvisé un poco allí, escalando los valores proyectados, pero si cree que sabe la respuesta, abra una edición.
Actualización 2: parece funcionar bien con la improvisación anterior
Stability.ai por el generoso patrocinio para trabajar en investigaciones de vanguardia en inteligencia artificial
einops por hacerme la vida más fácil
Triton por permitirme acelerar el descenso de coordenadas con una implementación fusionada en solo 2 días, ahorrándome tener que escribir mil líneas de código CUDA.
$ pip install colt5-attention import torch
from colt5_attention import (
ConditionalRoutedFeedForward ,
ConditionalRoutedAttention ,
ConditionalRoutedTransformerBlock
)
# mock input, say it is 32768 length
tokens = torch . randn ( 2 , 32768 , 512 )
mask = torch . ones ( 2 , 32768 ). bool () # can handle variable lengthed sequences
# feedforward
ff = ConditionalRoutedFeedForward (
dim = 512 ,
light_ff_mult = 0.5 , # hidden dimension ratio of light branch
heavy_ff_mult = 4 , # hidden dimension ratio of heavy branch
num_heavy_tokens = 1024 # heavy branch receives only 1024 routed tokens of 32768
)
ff_out = ff ( tokens , mask = mask ) # (2, 32768, 512) - light and heavy branch summed
# attention
attn = ConditionalRoutedAttention (
dim = 512 ,
light_dim_head = 64 , # attention head dimension of light branch
light_heads = 8 , # number of attention heads for light branch
light_window_size = 128 , # local attention receptive field for light
heavy_dim_head = 64 , # attention head dimension of heavy branch
heavy_heads = 8 , # number of attention heads for heavy branch
num_heavy_tokens_q = 1024 , # heavy branch receives only 1024 routed tokens of 32768
num_heavy_tokens_kv = 1024 # heavy branch receives only 1024 routed tokens of 32768
)
attn_out = attn ( tokens , mask = mask ) # (2, 32768, 512) - light and heavy branch summed
# both attention and feedforward with residual
# the complete transformer block
# a stack of these would constitute the encoder of CoLT5
block = ConditionalRoutedTransformerBlock (
dim = 512 ,
light_dim_head = 64 ,
light_heads = 8 ,
light_window_size = 128 ,
heavy_dim_head = 64 ,
heavy_heads = 8 ,
light_ff_mult = 0.5 ,
heavy_ff_mult = 4 ,
num_heavy_ff_tokens = 1024 ,
num_heavy_attn_tokens_q = 1024 ,
num_heavy_attn_tokens_kv = 1024
)
block_out = block ( tokens , mask = mask ) # (2, 32768, 512)También se incluye una variación de la atención enrutada condicionalmente para atención cruzada, que se probará con memorias de contexto largas en un transformador-xl.
import torch
from colt5_attention import ConditionalRoutedCrossAttention
# mock input, let us say it is a transformer of 1024 length attending to 1 million context past memories
tokens = torch . randn ( 1 , 1024 , 512 ). cuda ()
tokens_mask = torch . ones ( 1 , 1024 ). bool (). cuda ()
memories = torch . randn ( 1 , 1_048_576 , 512 ). cuda ()
memories_mask = torch . ones ( 1 , 1_048_576 ). bool (). cuda ()
# conditionally routed cross attention
cross_attn = ConditionalRoutedCrossAttention (
dim = 512 ,
dim_head = 64 ,
heads = 8 ,
num_tokens_q = 512 , # only 512 routed from 1024
num_tokens_kv = 1024 , # only 1024 routed from 1 million
kv_routing_tokens = 2 , # say you want 2 routing tokens to route different sets of key / values to the queries. 4 attention heads will be allocated to each routed set in this example (8 / 2)
use_triton = True , # use cuda kernel
route_block_size = 131072 # route in blocks of 131072
). cuda ()
cross_attn_out = cross_attn (
tokens ,
context = memories ,
mask = tokens_mask ,
context_mask = memories_mask
)
cross_attn_out . shape # (1, 1024, 512) - same as tokensEste repositorio también tiene una versión improvisada para atención autorregresiva. La forma en que esto se logró fue viendo la secuencia en Windows. Cada ventana solo puede atender a ventanas de claves/valores del pasado. La atención local de la rama de luz cubre la atención dentro de la ventana.
El descenso de coordenadas se hace viable mediante un kernel CUDA escrito en Triton. Finalmente, para que la generación autorregresiva funcione bien, tuve que asegurarme de que los tokens no enrutados (para consultas) generen una salida aprendida incrustada en lugar de solo ceros.
Actualmente veo diferencias ocasionales entre los gradientes (hasta 1e-1 para una fracción muy pequeña de elementos) una vez que el número de iteraciones supera las 20. Sin embargo, enwik8 parece funcionar bien y puedo ver los efectos del enrutamiento. El entrenamiento también es sorprendentemente estable
ex.
import torch
from colt5_attention import ConditionalRoutedAutoregressiveAttention
# mock input, say it is 8192 length
tokens = torch . randn ( 2 , 8192 , 512 ). cuda ()
# attention
attn = ConditionalRoutedAutoregressiveAttention (
dim = 512 ,
light_dim_head = 64 , # attention head dimension of light branch
light_heads = 8 , # number of attention heads for light branch
light_window_size = 128 , # local attention receptive field for light
heavy_window_size = 128 , # the windowing for the routed heavy attention, by default, will be equal to the light window size. be aware if this is any greater than the light window size, there may be tokens that would be missed by attention
heavy_dim_head = 64 , # attention head dimension of heavy branch
heavy_heads = 8 , # number of attention heads for heavy branch
num_heavy_tokens_q = 32 , # heavy branch receives only 32 out of 128 of the windowed queries (1024 query tokens total)
num_heavy_tokens_kv = 1024 , # heavy branch receives only 1024 routed tokens for key-values
num_routed_kv = 2 , # one can split the attention heads so that groups of heads attend to different sets of key - values (2 routing tokens in this case)
use_triton = True , # will need to use Triton for this to be viable, otherwise it is too slow and memory efficient with the number of iterations
use_flash_attn = True # use flash attention in heavy branch
). cuda ()
attn_out = attn ( tokens ) + tokens # (2, 8192, 512) - output of attention with residual (prenorm is included)Finalmente, este repositorio contiene una versión para mapas de características de imágenes. Por lo general, muchos trabajos de investigación no pueden prestar atención a mapas de características de imágenes con dimensiones superiores a 32 por 32. Esta atención enrutada utilizará un parche de ventana local para la rama liviana y atención enrutada para la rama pesada.
ex.
import torch
from colt5_attention import ConditionalRoutedImageAttention
attn = ConditionalRoutedImageAttention (
dim = 32 ,
light_dim_head = 64 , # attention head dimension of light branch
light_heads = 8 , # number of attention heads for light branch
light_window_size = 32 , # height and width of local window attention on the image feature map
channel_first = True , # whether to accept images with channel first than last
heavy_dim_head = 64 , # attention head dimension of heavy branch
heavy_heads = 8 , # number of attention heads for heavy branch
num_heavy_tokens_q = 1024 , # heavy branch receives only 1024 routed tokens of 65536
num_heavy_tokens_kv = 1024 # heavy branch receives only 1024 routed tokens of 65536
). cuda ()
fmap = torch . randn ( 1 , 32 , 256 , 256 ). cuda () # image feature map is too large for attention, given 256 ^ 2 == 65536 tokens
out = attn ( fmap )ViT simple que utiliza atención enrutada de descenso coordinado y retroalimentación
import torch
from colt5_attention . vit import ConditionalRoutedViT
vit = ConditionalRoutedViT (
image_size = 256 , # image size
patch_size = 32 , # patch size
num_classes = 1000 , # number of output classes
dim = 1024 , # feature dimension
depth = 6 , # depth
attn_num_heavy_tokens_q = 16 , # number of routed queries for heavy attention
attn_num_heavy_tokens_kv = 16 , # number of routed key/values for heavy attention
attn_heavy_dim_head = 64 , # dimension per attention head for heavy
attn_heavy_heads = 8 , # number of attention heads for heavy
attn_light_window_size = 4 , # the local windowed attention for light branch
attn_light_dim_head = 32 , # dimension per head for local light attention
attn_light_heads = 4 , # number of attention heads for local windowed attention
ff_num_heavy_tokens = 16 , # number of tokens routed for heavy feedforward
ff_heavy_mult = 4 , # the expansion factor of the heavy feedforward branch
ff_light_mult = 2 # expansion factor of the light feedforward branch
)
images = torch . randn ( 1 , 3 , 256 , 256 )
logits = vit ( images ) # (1, 1000) Utilice una pequeña envoltura alrededor del descenso de coordenadas para obtener topk diferenciable
import torch
from colt5_attention import topk
x = torch . randn ( 1024 , 512 )
values , indices , coor_descent_values , gates = topk ( x , k = 10 , fused = True )
# you can either use the topk indices + gates, or use the values directly (values have already been multiplied with the gates within the function) @inproceedings { Ainslie2023CoLT5FL ,
title = { CoLT5: Faster Long-Range Transformers with Conditional Computation } ,
author = { Joshua Ainslie and Tao Lei and Michiel de Jong and Santiago Ontan'on and Siddhartha Brahma and Yury Zemlyanskiy and David Uthus and Mandy Guo and James Lee-Thorp and Yi Tay and Yun-Hsuan Sung and Sumit Sanghai } ,
year = { 2023 }
} @article { Tillet2019TritonAI ,
title = { Triton: an intermediate language and compiler for tiled neural network computations } ,
author = { Philippe Tillet and H. Kung and D. Cox } ,
journal = { Proceedings of the 3rd ACM SIGPLAN International Workshop on Machine Learning and Programming Languages } ,
year = { 2019 }
} @inproceedings { dao2022flashattention ,
title = { Flash{A}ttention: Fast and Memory-Efficient Exact Attention with {IO}-Awareness } ,
author = { Dao, Tri and Fu, Daniel Y. and Ermon, Stefano and Rudra, Atri and R{'e}, Christopher } ,
booktitle = { Advances in Neural Information Processing Systems } ,
year = { 2022 }
} @article { Lei2023ConditionalAP ,
title = { Conditional Adapters: Parameter-efficient Transfer Learning with Fast Inference } ,
author = { Tao Lei and Junwen Bai and Siddhartha Brahma and Joshua Ainslie and Kenton Lee and Yanqi Zhou and Nan Du and Vincent Zhao and Yuexin Wu and Bo Li and Yu Zhang and Ming-Wei Chang } ,
journal = { ArXiv } ,
year = { 2023 } ,
volume = { abs/2304.04947 }
} @article { Beyer2022BetterPV ,
title = { Better plain ViT baselines for ImageNet-1k } ,
author = { Lucas Beyer and Xiaohua Zhai and Alexander Kolesnikov } ,
journal = { ArXiv } ,
year = { 2022 } ,
volume = { abs/2205.01580 }
}
