FileNotFoundError: [Errno 2] No such file or directory: 'data/src-train.embed'

[li3cmz]

Looking forward to your reply.

[gmftbyGMFTBY]

Sorry, I didn't show the tutorial for this repo.
Please run the utils.py to generate the model-ready data

python utils.py --mode dataset --dataset xxxx

Dataset's name should be the same as the folder's name under the folder data

I push the new commit and you can check it.
After processing the dataset, you can run the code by running

./run.sh $CUDA_DEVICE $dataset_name

It should be noted that the BERT-RUBER requires bert-as-service to generate the BERT contextual word embeddings.

I will update the README in a few days, and create a new repo which contains our proposed learning-based method (Better than RUBER and BERT-RUBER).

Thank you for your attention.

[gmftbyGMFTBY]

As for the datasets, you can check the following benchmarks:

• Xiaohuangji: https://github.com/fate233/dgk_lost_conv
• Tencent: arXiv 2018 / A Manually Annotated Chinese Corpus for Non-task-oriented Dialogue
• Dailydialog: https://github.com/Sanghoon94/DailyDialogue-Parser
• Cornell：http://www.cs.cornell.edu/~cristian/Cornell_Movie-Dialogs_Corpus.html

You need to process the dataset into the single-turn dialogue format, which contains 6 pure text file:

• src-train.txt: context or query of the conversation, one line one sentence
• tgt-train.txt: the ground-truth response for the query, one line one sentence
• src-test.txt
• tgt-test.txt
• src-dev.txt
• tgt-dev.txt

After running python utils.py --mode dataset --dataset xxxx, you can get the *.embed file which will be used for training the model.

[li3cmz]

Wow! Thank you for replying promptly! I will go to have a try!
And I will keep attention for your new repo. Thanks~

[ImmortalCi]

@gmftbyGMFTBY @li3cmz
Hello, I want to reproduce the result and deploy this model to accomplish unreferenced text generation evaluation. This is the topic of my thesis, could you provide processed datasets? Thanks a lot!