block recurrent transformer pytorch Download - block recurrent transformer pytorch Quellcode herunterladen

block recurrent transformer pytorch

AI-Quellcode

0.4.4

Herunterladen

Blockieren Sie wiederkehrende Transformatoren – Pytorch

Implementierung eines Block-Recurrent-Transformators – Pytorch. Der Höhepunkt des Papiers ist seine berichtete Fähigkeit, sich an etwas vor bis zu 60.000 Token zu erinnern.

Dieses Design ist SOTA für die Forschungslinie für wiederkehrende Transformatoren, afaict.

Es wird auch Flash-Aufmerksamkeit sowie geroutete Erinnerungen an bis zu 250.000 Token unter Verwendung von Ideen aus diesem Papier beinhalten

Anerkennung

Stability.ai für das großzügige Sponsoring der Arbeit und der Open-Source-Forschung im Bereich der künstlichen Intelligenz

Installieren

$ pip install block-recurrent-transformer-pytorch

Verwendung

 import torch
from block_recurrent_transformer_pytorch import BlockRecurrentTransformer

model = BlockRecurrentTransformer (
    num_tokens = 20000 ,             # vocab size
    dim = 512 ,                      # model dimensions
    depth = 6 ,                      # depth
    dim_head = 64 ,                  # attention head dimensions
    heads = 8 ,                      # number of attention heads
    max_seq_len = 1024 ,             # the total receptive field of the transformer, in the paper this was 2 * block size
    block_width = 512 ,              # block size - total receptive field is max_seq_len, 2 * block size in paper. the block furthest forwards becomes the new cached xl memories, which is a block size of 1 (please open an issue if i am wrong)
    num_state_vectors = 512 ,        # number of state vectors, i believe this was a single block size in the paper, but can be any amount
    recurrent_layers = ( 4 ,),        # where to place the recurrent layer(s) for states with fixed simple gating
    use_compressed_mem = False ,     # whether to use compressed memories of a single block width, from https://arxiv.org/abs/1911.05507
    compressed_mem_factor = 4 ,      # compression factor of compressed memories
    use_flash_attn = True           # use flash attention, if on pytorch 2.0
)

seq = torch . randint ( 0 , 2000 , ( 1 , 1024 ))

out , mems1 , states1 = model ( seq )
out , mems2 , states2 = model ( seq , xl_memories = mems1 , states = states1 )
out , mems3 , states3 = model ( seq , xl_memories = mems2 , states = states2 )

Test auf Enwik8

Zuerst pip install -r requirements.txt , dann

$ python train.py

Todo

Zitate

 @article { Hutchins2022BlockRecurrentT ,
    title   = { Block-Recurrent Transformers } ,
    author  = { DeLesley S. Hutchins and Imanol Schlag and Yuhuai Wu and Ethan Dyer and Behnam Neyshabur } ,
    journal = { ArXiv } ,
    year    = { 2022 } ,
    volume  = { abs/2203.07852 }
}

 @article { Shazeer2019FastTD ,
    title   = { Fast Transformer Decoding: One Write-Head is All You Need } ,
    author  = { Noam M. Shazeer } ,
    journal = { ArXiv } ,
    year    = { 2019 } ,
    volume  = { abs/1911.02150 }
}

 @inproceedings { Sun2022ALT ,
    title     = { A Length-Extrapolatable Transformer } ,
    author    = { Yutao Sun and Li Dong and Barun Patra and Shuming Ma and Shaohan Huang and Alon Benhaim and Vishrav Chaudhary and Xia Song and Furu Wei } ,
    year      = { 2022 }
}

 @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 }
}

 @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 }
}