Rotary Embeddings - Pytorch
A standalone library for adding rotary embeddings to transformers in Pytorch, following its success as relative positional encoding. Specifically it will make rotating information into any axis of a tensor easy and efficient, whether they be fixed positional or learned. This library will give you state of the art results for positional embedding, at little costs.
My gut also tells me there is something more to rotations that can be exploited in artificial neural networks.
Install
$ pip install rotary-embedding-torch
Usage
import torch
from rotary_embedding_torch import apply_rotary_emb, RotaryEmbedding
# instantiate the positional embedding in your transformer and pass to all your attention layers
pos_emb = RotaryEmbedding(dim = 32)
# generate the rotations
freqs = pos_emb(torch.arange(1024), cache_key = 1024) # cache with a key that is the sequence length, so that it does not need to recompute
# mock queries and keys
q = torch.randn(1, 1024, 64) # queries - (batch, seq len, dimension of head)
k = torch.randn(1, 1024, 64) # keys
# apply the rotations to your queries and keys after the heads have been split out, but prior to the dot product and subsequent softmax (attention)
freqs = freqs[None, ...] # unsqueeze for batch dimension
q = apply_rotary_emb(freqs, q)
k = apply_rotary_emb(freqs, k)
# then do your attention with your queries (q) and keys (k)
If you do all the steps above correctly, you should see a dramatic improvement during training
Axial Rotary Embeddings
For easy use of 2d axial relative positional embedding, ie. vision transformers
import torch
from rotary_embedding_torch import apply_rotary_emb, RotaryEmbedding, broadcat
pos_emb = RotaryEmbedding(
dim = 32,
freqs_for = 'pixel'
)
# queries and keys for frequencies to be rotated into
q = torch.randn(1, 256, 256, 64)
k = torch.randn(1, 256, 256, 64)
# get frequencies for each axial
# -1 to 1 has been shown to be a good choice for images and audio
freqs_h = pos_emb(torch.linspace(-1, 1, steps = 256), cache_key = 256)
freqs_w = pos_emb(torch.linspace(-1, 1, steps = 256), cache_key = 256)
# concat the frequencies along each axial
# broadcat function makes this easy without a bunch of expands
freqs = broadcat((freqs_h[None, :, None, :], freqs_w[None, None, :, :]), dim = -1)
# rotate in frequencies
q = apply_rotary_emb(freqs, q)
k = apply_rotary_emb(freqs, k)
Learned Rotations
For injecting learned rotations into a network. Experiments pending
Update: doesn't seem to do anything -_-, will keep trying...
import torch
from torch import nn
from rotary_embedding_torch import apply_learned_rotations
x = torch.randn(1, 1024, 512)
# you can only rotate in (dim // 2) values
# ex. for 512, you can only rotate in 256 values
# say you have two sets of learned rotations of 128 values each
rots1 = nn.Linear(512, 128)(x)
rots2 = nn.Linear(512, 128)(x)
# you rotate in 256 (128 x 2) at first
x = apply_learned_rotations(rots1, x, start_index = 0)
# then you start at index 256 and rotate in the last (128 x 2)
x = apply_learned_rotations(rots2, x, start_index = 256)
# you could also concat the rotations together and pass it in all at once
rots = torch.cat((rots1, rots2), dim = -1)
x = apply_learned_rotations(rots, x)
Citations
@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}
}
253 Dec 18, 2022
3.8k Dec 22, 2022
132 Dec 13, 2022
242 Dec 30, 2022
88 Dec 30, 2022
34 Dec 31, 2022
71 Oct 25, 2022
1.1k Jan 9, 2023
107 Jan 3, 2023
58 Nov 30, 2022
13 Jun 28, 2022
594 Jan 08, 2023
291 Dec 24, 2022
2 May 19, 2022
300 Dec 28, 2022
142 Dec 15, 2022
42 Oct 29, 2022
10 Dec 17, 2022
6 Dec 25, 2022
35 Nov 24, 2022
460 Jan 05, 2023
2 Apr 18, 2022
8 May 11, 2022
5 Oct 30, 2022
175 Jan 04, 2023
23 Nov 24, 2022
431 Dec 27, 2022
105 Dec 25, 2022