phizaz

import os
import random

import cv2
import pandas as pd
import torch
from torch.utils.data import Dataset
from transformers import (BertTokenizerFast, PreTrainedTokenizerFast,
                          RobertaTokenizerFast)

phizaz

from segmentation_models_pytorch.encoders import get_encoder

from .mil import MILPool

def make_net_li2018(
    backbone,
    n_out,
    n_in=1,
    n_dec_ch=512,
    out_size=20,

phizaz

import errno
import os
import time
from collections import defaultdict
from contextlib import ContextDecorator, contextmanager
from itertools import count

import torch

CUDA_ALLOC_FILE = os.path.expanduser('~/mlkit.alloc')

phizaz

"""
blocks until a specific time, 
useful for planning to run a script at a specific time

usage: 

python til.py 13:00 tomorrow && python script.py
"""
import dateparser
from datetime import datetime, timedelta

phizaz

phizaz

import tensorflow as tf
import numpy as np

tf.enable_eager_execution()

save_path = 'logs/test10'
graph = tf.Graph()
with graph.as_default():
    global_step = tf.train.create_global_step()
    writer = tf.contrib.summary.create_file_writer(save_path)

phizaz

import tensorflow as tf
import numpy as np

save_path = 'summary_path'
graph = tf.Graph()
with graph.as_default():
    global_step = tf.train.create_global_step()
    writer = tf.contrib.summary.create_file_writer(save_path)
    with writer.as_default():
        tf.contrib.summary.always_record_summaries()

phizaz

import tensorflow.contrib.summary as summary

global_step = tf.train.get_or_create_global_step()
optimizer = tf.train.AdamOptimizer(lr).minimize(..., global_step=global_step)

writer = summary.create_file_writer('log')
writer.set_as_default()
summary.always_record_summaries()

with summary.record_summaries_every_n_global_steps(1):

phizaz

from select import select
import os
import time
import subprocess
from contextlib import contextmanager


@contextmanager
def pipe():
    r, w = os.pipe()

phizaz

class MyConv(Layer):
    '''
    Implemeting a Conv2D with strides=1, and 'valid' padding
    '''
    
    def __init__(self, filters, kernel, **kwargs):
        self.filters = filters
        self.k_h, self.k_w = kernel
        super(MyConv, self).__init__(**kwargs)