Retrain with up to itr 23
juanmirocks opened this issue · comments
- Add itr18 & itr19 reviewed
- Verify why, currently, before itr18 & itr19, itr16 to 21 are not counted in read_learning_data nor in check or print stacks. Below the current report of check stats. As a result, currently only 298 docs are counted, whereas in fact we have 377
--> Explained by Alex, check in place anncomplete==true - Make anncomplete a parameter of annjson readers
- Make use all training data available until the moment
- Train model
- Share model (tell @jmcejuela to put on public ftp)
--> ftp://rostlab.org/jmcejuela/
####ReadIterationData####
iter | 15.00
total | 30.00
st | 9.00
nl + ss | 21.00
nl | 18.00
ss | 3.00
nl+ss/doc | 2.10
####ReadIterationData####
iter | 16.00
total | 0.00
st | 0.00
nl + ss | 0.00
nl | 0.00
ss | 0.00
nl+ss/doc | 0.00
####ReadIterationData####
iter | 17.00
total | 0.00
st | 0.00
nl + ss | 0.00
nl | 0.00
ss | 0.00
nl+ss/doc | 0.00
####ReadIterationData####
iter | 18.00
total | 0.00
st | 0.00
nl + ss | 0.00
nl | 0.00
ss | 0.00
nl+ss/doc | 0.00
####ReadIterationData####
iter | 19.00
total | 0.00
st | 0.00
nl + ss | 0.00
nl | 0.00
ss | 0.00
nl+ss/doc | 0.00
####ReadIterationData####
iter | 20.00
total | 0.00
st | 0.00
nl + ss | 0.00
nl | 0.00
ss | 0.00
nl+ss/doc | 0.00
####ReadIterationData####
iter | 21.00
total | 0.00
st | 0.00
nl + ss | 0.00
nl | 0.00
ss | 0.00
nl+ss/doc | 0.00
***** Iteration #1824 *****
Loss: 1492.277583
Feature norm: 28.222768
Error norm: 14.341202
Active features: 207897
Line search trials: 1
Line search step: 1.000000
Seconds required for this iteration: 2.088
L-BFGS terminated with the stopping criteria
Total seconds required for training: 4387.969
Storing the model
Number of active features: 207897 (207897)
Number of active attributes: 172956 (172956)
Number of active labels: 4 (4)
Writing labels
Writing attributes
Writing feature references for transitions
Writing feature references for attributes
Seconds required: 1.454