How to get sent_sg_parse?

[cici-ai-club]

Hi, wentian, I am very interested in MemCap. I would like to get the sent_sg_parse since use_sg is True for the train. But I failed to find where and how to generate based on the code. I am thinking it is the part called "Stylized Sentence Decomposing" in the paper. Any implementation on how to generate that?

[cici-ai-club]

I'm also waiting for wentian's help~do you successfully to preprocess the senticap datasets? I cannot find any code about the part.

Not really. That part is not very clear yet. But I do see the Vocab and preprocess code files. It might help for that. It will be awesome if we can get the preprocess from wentian. Temporarily cannot have a successful run through because of lacking files. Since decomposing is essential part of the paper, I actually would like to play with it.

[n9705]

Is this code written by the author of the paper？The lack of README make me comfuse about the authority of this code.

[cici-ai-club]

Is this code written by the author of the paper？The lack of README make me comfuse about the authority of this code.

i think the code is truely written by the author of the paper !
you can see config.py line 19
kenlm_build_path = '/home/wentian/library/kenlm/build'
wentian is the author of the paper.

Yeah, I agree, at least wentian provide something rather than nothing. Also, I hope more details if he or she had time.

[cici-ai-club]

Is this code written by the author of the paper？The lack of README make me comfuse about the authority of this code.

i think the code is truely written by the author of the paper !
you can see config.py line 19
kenlm_build_path = '/home/wentian/library/kenlm/build'
wentian is the author of the paper.

Yeah, I agree, at least wentian provide something rather than nothing. Also, I hope more details if he or she had time.

Though something provided, i cannot run the code simply. 😢 How about you? can you reproduce the result successfully？

I can train the style classifier and language model he provided. I think that is very helpful for me. Hope you could get something too.

[qin79]

Is this code written by the author of the paper？The lack of README make me comfuse about the authority of this code.

i think the code is truely written by the author of the paper !
you can see config.py line 19
kenlm_build_path = '/home/wentian/library/kenlm/build'
wentian is the author of the paper.

Yeah, I agree, at least wentian provide something rather than nothing. Also, I hope more details if he or she had time.

Though something provided, i cannot run the code simply. 😢 How about you? can you reproduce the result successfully？

I can train the style classifier and language model he provided. I think that is very helpful for me. Hope you could get something too.

would you please give some guidance on running the code？thanks a lot！

[cici-ai-club]

Is this code written by the author of the paper？The lack of README make me comfuse about the authority of this code.

i think the code is truely written by the author of the paper !
you can see config.py line 19
kenlm_build_path = '/home/wentian/library/kenlm/build'
wentian is the author of the paper.

Yeah, I agree, at least wentian provide something rather than nothing. Also, I hope more details if he or she had time.

Though something provided, i cannot run the code simply. 😢 How about you? can you reproduce the result successfully？

I can train the style classifier and language model he provided. I think that is very helpful for me. Hope you could get something too.

would you please give some guidance on running the code？thanks a lot！

I hope the author could get you back cause I am not the author of this paper and I only get hints of how they perform style evaluation in the styled_eval directory.