File size: 85 Bytes
5fa1a76
1
A transformer model replacing the attention matrices by sparse matrices to go faster.