Skip to content

Instantly share code, notes, and snippets.

View horiacristescu's full-sized avatar

Horia Cristescu horiacristescu

  • Bucharest
View GitHub Profile
@cbaziotis
cbaziotis / Attention.py
Last active March 28, 2023 11:50
Keras Layer that implements an Attention mechanism for temporal data. Supports Masking. Follows the work of Raffel et al. [https://arxiv.org/abs/1512.08756]
from keras import backend as K, initializers, regularizers, constraints
from keras.engine.topology import Layer
def dot_product(x, kernel):
"""
Wrapper for dot product operation, in order to be compatible with both
Theano and Tensorflow
Args:
@num3ric
num3ric / seqalignment.py
Created September 16, 2011 18:26
Sequence alignment - Dynamic programming algorithm
#!/usr/bin/python -O
import numpy as np
from numpy import array
A, C, G, T = 0, 1, 2, 3
int_to_char = {0:'A', 1:'C', 2:'G', 3:'T'}
#indel = -1
#scoring = array([[1,-1,-1,-1],
#[-1,1,-1,-1],