Implementation of MEGABYTE, Predicting Million-byte Sequences with Multiscale Transformers, in Pytorch. Took the liberty to generalize it even further so one can have multiple local models.
Similar independent research that is a further generalization
$ pip install MEGABYTE-pytorch
import torch
from MEGABYTE_pytorch import MEGABYTE
model = MEGABYTE(
num_tokens = 16000, # number of tokens
dim = (512, 256), # transformer model dimension (512 for coarsest, 256 for fine in this example)
max_seq_len = (1024, 4), # sequence length for global and then local. this can be more than 2
depth = (6, 4), # number of layers for global and then local. this can be more than 2, but length must match the max_seq_len's
dim_head = 64, # dimension per head
heads = 8, # number of attention heads
flash_attn = True # use flash attention
)
x = torch.randint(0, 16000, (1, 1024, 4))
loss = model(x, return_loss = True)
loss.backward()
# then after much training
logits = model(x)
# and sample from the logits accordingly
# or you can use the generate function
sampled = model.generate(temperature = 0.9, filter_thres = 0.9) # (1, 1024, 4)
Train on character-level enwik8 with patches of size 4 - length 8192
$ python train.py
@misc{yu2023megabyte,
title = {MEGABYTE: Predicting Million-byte Sequences with Multiscale Transformers},
author = {Lili Yu and Dániel Simig and Colin Flaherty and Armen Aghajanyan and Luke Zettlemoyer and Mike Lewis},
year = {2023},
eprint = {2305.07185},
archivePrefix = {arXiv},
primaryClass = {cs.LG}
}
@misc{https://doi.org/10.48550/arxiv.2302.01327,
doi = {10.48550/ARXIV.2302.01327},
url = {https://arxiv.org/abs/2302.01327},
author = {Kumar, Manoj and Dehghani, Mostafa and Houlsby, Neil},
title = {Dual PatchNorm},
publisher = {arXiv},
year = {2023},
copyright = {Creative Commons Attribution 4.0 International}
}
@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}
}
@software{peng_bo_2021_5196578,
author = {PENG Bo},
title = {BlinkDL/RWKV-LM: 0.01},
month = {aug},
year = {2021},
publisher = {Zenodo},
version = {0.01},
doi = {10.5281/zenodo.5196578},
url = {https://doi.org/10.5281/zenodo.5196578}
}
@article{Kazemnejad2023TheIO,
title = {The Impact of Positional Encoding on Length Generalization in Transformers},
author = {Amirhossein Kazemnejad and Inkit Padhi and Karthikeyan Natesan Ramamurthy and Payel Das and Siva Reddy},
journal = {ArXiv},
year = {2023},
volume = {abs/2305.19466}
}
@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}
}