“Qixin Wang 2016-01-25”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
(以“=== Work done in this week === word vector size:200 hidden size:500 mlp hidden size:400 maxout size:300 adadelta 0.3 fast mode, added cut, no global, no pz zgt:s...”为内容创建页面)
 
Wqx讨论 | 贡献
第3行: 第3行:
  
 
word vector size:200
 
word vector size:200
 +
 
hidden size:500
 
hidden size:500
 +
 
mlp hidden size:400
 
mlp hidden size:400
 +
 
maxout size:300
 
maxout size:300
 +
 
adadelta 0.3
 
adadelta 0.3
 +
 +
---
  
 
fast mode, added cut, no global, no pz
 
fast mode, added cut, no global, no pz
  
zgt:song, si, giga, update:
+
zgt:song, si, giga, update: (grid-9, grid-9, grid-17, grid-17)
  
grid-9, grid-9, grid-17, grid-17
+
psm:song, si, giga, update: (grid-15, grid-15, grid-13, grid-11)
  
psm:song, si, giga, update:
 
  
grid-15, grid-15, grid-13, grid-11
+
with dropout & without maxout:
  
 
batch_all(zgt): grid-12
 
batch_all(zgt): grid-12
  
 
batch_all_go(zgt): grid-11
 
batch_all_go(zgt): grid-11

2016年1月21日 (四) 00:54的版本

Work done in this week

word vector size:200

hidden size:500

mlp hidden size:400

maxout size:300

adadelta 0.3

---

fast mode, added cut, no global, no pz

zgt:song, si, giga, update: (grid-9, grid-9, grid-17, grid-17)

psm:song, si, giga, update: (grid-15, grid-15, grid-13, grid-11)


with dropout & without maxout:

batch_all(zgt): grid-12

batch_all_go(zgt): grid-11