Parlai: Confusions in the Blended Skill Talk dataset

[wanghaoran-ucas]

I don't understand the details of these datasets, Blended Skill Talk, Wizard of Wikipedia, Empathetic Dialog, ConvAI2 from parlAI (https://parl.ai/projects/bst/).

I have read the paper on each of these datasets, but the papers don't refer to column definition or file definition while they describe the kind or purpose of dataset.
Can someone tell me more about these datasets?

In particular, my question is:
I don't understand what the following files are. How can I use them?

(Blended Skill Talk)
"ed_persona_topicifier__test__experiencer_only.json"
"ed_persona_topicifier__train__both_sides.json"
"ed_persona_topicifier__train__experiencer_only.json"
"ed_persona_topicifier__valid__experiencer_only.json"
"persona_list.txt"
"safe_personas.txt"
"topic_to_persona_list.txt"
"train.txt"

(Wizard of Wikipedia)
"topic_splits.json"
"test_topic_split.json"
"valid_topic_split.json"

Are there any documents about the definition of these datasets?

[stephenroller]

All the datasets have teachers that load them. You can see, for example, by running:

python -m parlai.scripts.display_data -t convai2

I recommend you read our quickstart tutorial.

[llStringll]

@wanghaoran-ucas Refer the readme.md of this directory https://github.com/facebookresearch/ParlAI/tree/master/parlai/tasks/blended_skill_talk, it's very clear there, I read it

[wanghaoran-ucas]

thank you @llStringll @stephenroller

[github-actions[bot]]

This issue has not had activity in 30 days. Please feel free to reopen if you have more issues. You may apply the "never-stale" tag to prevent this from happening.