Lianhui Qin

The model should be able to learn generating EOS. This [line](https://github.com/qkaren/Counterfactual-StoryRW/blob/d16f8fb5ac7b829ad39d6c2afec4ab83dd1c0975/rewriting_main.py#L674) truncates all tokens after EOS. You may want to make sure that at training time the model does see...

> Hi @qkaren > I ran the code, straight from this repo, on the timetravel dataset and have the same issue. > The model rewrites the original story and then...

Sorry for the delay. It's likely that this version of code has a bug so that the model does not see the '' token at training time. I'm still trying...

hi Could you go through the README and see if you can get the data? https://github.com/qkaren/Counterfactual-StoryRW/blob/5e138d4ad11dd5d1d21dc20d869ffae594201734/README.md Thanks, Lianhui On Wed, Oct 13, 2021 at 8:59 AM ccyun1127 ***@***.***> wrote: >...

ok something is missing in this repo. I think you just need to create some files which are aligned by lines. train_x1.txt: premise train_x2.txt: initial train_xx2.txt: counterfactual train_y.txt: original ending...

oh yes, the edited_endings are used for multiple references evaluation. You may want to use all of them On Sun, Oct 17, 2021 at 2:51 AM ccyun1127 ***@***.***> wrote: >...

Hi Yeonchan, Did you try to follow this page? https://github.com/qkaren/converse_reading_cmr/tree/master/evaluation

oh for those grounding metrics, we probably didn't include them here. They are just the normal Precision and Recall methods that should be easy to implement.

Yes, for tokenizer and stemmer: ``` from nltk.tokenize import TweetTokenizer from nltk.stem.porter import * ``` for the stop word set: [stopwords_700+.txt](https://github.com/qkaren/converse_reading_cmr/files/6438307/stopwords_700%2B.txt)

``` def count_grounded(facts, fresult, ids, writer, count_examples,file): g_count = 0 w_count = 0 f_count = 0 lines_count = 0 fact_dict = dict() stop_words = _get_stop_words() print(len(facts), len(fresult)) assert len(facts) ==...