11 2020 档案

TF代码片段
摘要：* keys_len 如果是None的时候， 注释代码失败。 ```python # keys_len = keys.get_shape()[1] # queries = K.repeat_elements(query, keys_len, 1) keys_len = tf.shape(keys)[ 阅读全文
posted @ 2020-11-18 15:02 bregman 阅读(278) 评论(0) 推荐(0) 编辑
TF的测试例子
摘要：稀疏矩阵(或三维以上稀疏张量)的测试例子 import tensorflow as tf import numpy as np sess = tf.InteractiveSession() # 稀疏tensor构造 import numpy as np A = tf.constant(np.rand 阅读全文
posted @ 2020-11-12 20:33 bregman 阅读(599) 评论(0) 推荐(0) 编辑
TF的FLAGS字典获取
摘要：```python tf.app.flags.DEFINE_integer('num_blocks', 1, 'Number of blocks in each attention') tf.app.flags.DEFINE_integer('num_heads', 8, 'Number of he 阅读全文
posted @ 2020-11-09 19:58 bregman 阅读(135) 评论(0) 推荐(0) 编辑
TF中的自定义正则项
摘要：```python # 计算正则项 # t_l2_reg = tf.nn.l2_loss(emb, name=reg_name) * l2_reg t_l2_reg = tf.reduce_mean(tf.reduce_sum(emb**2, axis=-1), name=reg_name) * l 阅读全文
posted @ 2020-11-05 15:52 bregman 阅读(169) 评论(0) 推荐(0) 编辑

点击右上角即可分享
微信分享提示