Hi, thanks for sharing your code. I evaluate the training results, the prediction results obtained using eval_simpletod.py are very different from the results in the paper, can you provide the evaluation code corresponding to the results in the paper?

eval_simpletod.py：
Belief acc: 0.06394596924845523
Action acc: 0.2207213680126455
Inform acc: 0.8195142980313264
Request acc: 0.7255352780571921

In paper:
joint acc: 0.28
inform rate: 14.6
success rate: 9.2

Looking forward to your reply.

Hi, This work is of great help to me.

I can't find the image data that these ids are shown in the dialogue.json.
Can release the corresponding the image dataset ?

Thanks very much!!! Looking forward to your reply！

python generate_inputs.py

Loading image_classes.json...
Loaded image_classes.json to <class 'dict'> object
Loading ../dataset/evidence_for_delex.json...
Traceback (most recent call last):
File "generate_inputs.py", line 27, in
data = load(path + 'evidence_for_delex.json')
File "~/MMConv/MMDST/utils/json_utils.py", line 41, in load
with open(path) as f:
FileNotFoundError: [Errno 2] No such file or directory: '../dataset/evidence_for_delex.json'