Hi, I would like to clarify if this is a bug.

Should the extractSeq method in line 878 be called on turn['usr']['transcript'] instead?

The words variable in line 877 is totally ignored, and replaced by the return values of extractSeq in line 878.

I was wondering why you chose Thanos to base your implementation on? Instead of a more generic platform like PyTorch or Tensor flow.

Were there hard blockers with these platforms?
Because I imagine you could have leveraged a lot of things from it? Like all the NN operators and optimizers. Without having to implement them from scratch?

I found two bugs:

1.in line 377,"if v=='dontcare' and s=='this':",where s=='this' ,in CamRest676.json it's not found that the slot equal 'this'
issue code location:

2.in line 380,"if sda['act']=='request':",where sda['act'] ,variable sda has no attribute 'act' ,because in CamRest676.json,items of DA is string type and similar to the following:
"sys": { "sent": "There are several restaurants in the south part of town that serve expensive food. Do you have a cuisine preference?", "DA": [ "food" ] }
issue code location:

I have some problems:
when I use
python nndial.py -config config/tracker.cfg -mode train
it prompt me:
loading model settings from config file ...
prepare slot value templates ...
formatting DB ...
setting up vocab, finishing ... 99.85%
semi-supervised action examples: 0.00%
loading semi labels from file ... finishing 100.00%
Corpus VMC : 97.34%
Corpus Success : 91.57%

===============
Data statistics

Train : 405
Valid : 135
Test : 136

Voc : 598

Venue : 68

setting network structures using theano variables ...
init n2n SDS ...
init rnn requestable trackers ...
init OfferChange tracker ...
init rnn informable trackers ...
init normal policy network ...
loss function
/home/andysun100800/anaconda2/envs/nndial/lib/python2.7/site-packages/pygpu/init.py:7: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
from . import gpuarray, elemwise, reduction
/home/andysun100800/anaconda2/envs/nndial/lib/python2.7/site-packages/pygpu/elemwise.py:5: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
from ._elemwise import GpuElemwise, arg
/home/andysun100800/anaconda2/envs/nndial/lib/python2.7/site-packages/theano/gpuarray/blas.py:19: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
from pygpu import blas
/home/andysun100800/nndial/NNDIAL/nn/encoder.py:111: UserWarning: DEPRECATION: the 'ds' parameter is not going to exist anymore as it is going to be replaced by the parameter 'ws'.
ignore_border=False)
/home/andysun100800/nndial/NNDIAL/nn/tracker.py:98: UserWarning: DEPRECATION: If x is a vector, Softmax will not automatically pad x anymore in next releases. If you need it, pleasedo it manually. The vector case is gonna be supported soon and the output will be a vector.
b_j = T.nnet.softmax( g_j )[0,:]
including informable tracker loss ...
including informable tracker loss ...
including informable tracker loss ...
/home/andysun100800/nndial/NNDIAL/nn/tracker.py:222: UserWarning: DEPRECATION: If x is a vector, Softmax will not automatically pad x anymore in next releases. If you need it, please do it manually. The vector case is gonna be supported soon and the output will be a vector.
b_j = T.nnet.softmax( g_j )[0,:]
including requestable tracker loss ...
including requestable tracker loss ...
including requestable tracker loss ...
including requestable tracker loss ...
including requestable tracker loss ...
including requestable tracker loss ...
including OfferChange tracker loss ...
gradient w.r.t inftrk
gradient w.r.t reqtrk

how to solve this problem?

I use default config and run the tracker training on macOS:
python nndial.py -config config/tracker.cfg -mode train

logs below：

init net from scrach ...
loading model settings from config file ...
prepare slot value templates ...
formatting DB ...
semi-supervised action examples: 0.00%
Corpus VMC : 97.34%
Corpus Success : 91.57%
===============
Data statistics
===============
Train : 405
Valid : 135
Test : 136
===============
Voc : 598
===============
Venue : 68
===============
setting network structures using theano variables ...
init n2n SDS ...
init rnn requestable trackers ...
init OfferChange tracker ...
init rnn informable trackers ...
init normal policy network ...
loss function
including informable tracker loss ...
including informable tracker loss ...
including informable tracker loss ...
including requestable tracker loss ...
including requestable tracker loss ...
including requestable tracker loss ...
including requestable tracker loss ...
including requestable tracker loss ...
including requestable tracker loss ...
including OfferChange tracker loss ...
gradient w.r.t inftrk
gradient w.r.t reqtrk

issue：
Program is blocked here,the log is no longer printed.
Apple Activity Monitor status:CPU is 98%,Memory is 15.56GB.

hi, I am reading your code .
I think you are using only one cnn kernel in each layer for belief tracking ,am I right?
can you explain for me ,thank you

I am training the belief tracker. I am getting the error in the very last stage of training:

.
.
.
Finishing 202 dialog in epoch 1
Finishing 203 dialog in epoch 1
Epoch 1, Alpha 0.002000, TRAIN entropy:95.17, Time:14.06 mins,
VALID entropy:74.52
saving net to file ...
Traceback (most recent call last):
File "nndial.py", line 22, in
model.trainNet()
File "C:\Users\MY DELL\Desktop\Course\6th sem\UGP\data\NNDIAL-master\nn\NNDialogue.py", line 590, in trainNet
self.saveNet()
File "C:\Users\MY DELL\Desktop\Course\6th sem\UGP\data\NNDIAL-master\nn\NNDialogue.py", line 999, in saveNet
'file' :dict( [(name,eval(name)) for name in self.file_vars]),
File "C:\Users\MY DELL\Desktop\Course\6th sem\UGP\data\NNDIAL-master\nn\NNDialogue.py", line 999, in
'file' :dict( [(name,eval(name)) for name in self.file_vars]),
File "", line 1, in
NameError: name 'self' is not defined

I am not able to understand why this error is coming

Could anyone please help with this. I would be really grateful. Thanks

LIDM_70
Venue Match Rate : 91.9%
Task Success Rate : 74.3%
BLEU : 0.2337
Semantic Match : 63.7%

LIDM-RL_70
Venue Match Rate : 92.6%
Task Success Rate : 75.7%
BLEU : 0.2459
Semantic Match : 63.4%

LIDM-RL_100 (just set I = 100)
Venue Match Rate : 89.7%
Task Success Rate : 70.6%
BLEU : 0.2541
Semantic Match : 60.3%

not like the paper's result:
LIDM Models
LIDM, I = 50 66.9 0.238
LIDM, I = 70 61.0 0.246
LIDM, I = 100 63.2 0.242
LIDM Models + RL
LIDM, I = 50, +RL 82.4 0.231
LIDM, I = 70, +RL 81.6 0.230
LIDM, I = 100, +RL 84.6 0.240

shawnwun, thanks.

try run :
python scp/batch_script/train_LIDM.py

if comment out the line including hpc_submit, then
Traceback (most recent call last):
File "scp/batch_script/train_LIDM.py", line 58, in
fout.write( configN2N(arc['enc'],arc['dec'],arc['trk']) )
NameError: name 'arc' is not defined