atc-torch 0.0.1

atc-torch - Pytorch

Attention + Convolution transformer

This is an experimental architecture leveraging convolution blocks with attention blocks to model both the short and long range dynamics of the input tokens. The flow is the following: x -> convolution block -> attn -> FFN

Install

``

Usage

import torch
from attnconv.main import ATCTransformer

model = ATCTransformer(
    dim=512,
    depth=6,
    num_tokens=20000,
    dim_head=64,
    heads=8,
    ff_mult=4,
)

x = torch.randint(0, 20000, (1, 512))
logits = model(x)  # (1, 1024, 20000)
print(logits)

License

MIT