# sgd
w1.assign_sub(learning_rate * grads[0])
b1.assign_sub(learning_rate * grads[1])
# sgd-momentun beta = 0.9
m_w = beta * m_w + (1 - beta) * grads[0]
m_b = beta * m_b + (1 - beta) * grads[1]
w...
| English Article | Word Count | Duration | Accuracy | Speed | Backspace | Error | Slogan |
|---|---|---|---|---|---|---|---|
| 《TensorFlow第二讲》 | 4584字 | 60.0分钟 | 91% | 82 CPM | 410 次 | 399 | 不及格，错误太多！ |
| The wrong word this time: #,s,g,d,指,数,衰,减,学,习,率,策,略,.,分,段,常,数,衰,减,学,习,率,策,略,.,,激,活,函,数,计,算,x,每,一,个,元,素,的,s,i,g,m,o,i,d,值,.,计,算,x,每,一,个,元,素,的,双,曲,正,切,值,.,#,等,价,实,现,计,算,L,e,a,k,y,R,e,L,U,值,.,与,l,o,g,i,t,s,s,h,a,p,e,相,同,的,张,量,.,#,等,价,实,现,损,失,函,数,#,等,价,实,现,计,算,交,叉,熵,.,#,等,价,实,现,l,o,g,i,t,s,经,过,s,o,f,t,m,a,x,后,,,与,l,a,b,e,l,s,进,行,交,叉,熵,计,算,.,#,等,价,实,现,功,能,:,l,a,b,e,l,s,经,过,o,n,e,-,h,o,t,编,码,,,l,o,g,i,t,s,经,过,s,o,f,t,m,a,x,,,两,者,进,行,交,叉,熵,计,算,.,通,常,l,a,b,e,l,s,的,s,h,a,p,e,为,[,b,a,t,c,h,_,s,i,z,e,],,,l,o,g,i,t,s,的,s,h,a,p,e,为,[ | |||||||
|
Suggestion:
正确率比较低，多观察一下出错的字加以修正，一般来讲都要达到95%以上比较好。
可能对按键不够熟悉，或者对输入法不够熟练所导致，不要急慢慢来，先打对把准确率提高，然后逐渐的加快速度，最后才能打准打快。
|
|||||||
| Inspirational quotes: 不为时尚所惑，不为积习所蔽，不为浮名所累。 ——乔羽 | |||||||
| What do you think of this article? Give it a rating and let it serve as a guiding light for others | |||||||
Registration with the Ministry of Industry and Information Technology: Min ICP No. 19000005