This repository has been archived by the owner on Nov 3, 2023. It is now read-only.
[TOD][Dataset][Easyish] Google Simulation Splits #4182
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
added 16 commits
November 15, 2021 20:15
See documentation block in `tod_agents.py`
As noted in the README, this agent takes data generated from `tod_world_script.py` and dumps it out to a teacher. (Note that I tried setting up a regression test for this teacher, but I ran into issues getting it to save the output directory to not be something that included my local homedir name in it..)
See documentation block in `tod_agents.py` (I'm not 100% sure if `conftest.py` is a right file to change, though I did notice that `pytest.ini` was necessary to get pytest to run.)
See documentation in `tod_world_script.py` for usage.
…iff in stack, but lost from this one
Refactor Google SGD away from old format into TOD Conversations format. Datasets added in this substack: * *Google SGD* * Google SGD Simulation Splits (In-domain, Out-domain) * MetalWoz * MSR_E2E * Multidogo * MultiWoz V2.2 * Taskmaster * Taskmaster2 * Taskmaster3 (TicketTalk) Test plan: Regression test, `parlai dd` of dataset
Code for processing Google SGD into In-domain and Out-domain data via `build.py`, using via agents. Datasets added in this substack: * Google SGD * **Google SGD Simulation Splits (In-domain, Out-domain)** * MetalWoz * MSR_E2E * Multidogo * MultiWoz V2.2 * Taskmaster * Taskmaster2 * Taskmaster3 (TicketTalk) Test plan: Regression test, `parlai dd` of dataset
moyapchen
requested review from
stephenroller,
chinnadhurai,
skiingpacman and
pooyanamini
added 9 commits
November 16, 2021 14:47
added 17 commits
November 30, 2021 09:46
…mplications for mutators too?)
stephenroller
approved these changes
Dec 12, 2021
added 6 commits
December 17, 2021 21:49
…ta and new one; relized I was missing a +1 in the episode length count
|
Ran an eval on the new dataset with the old models; get the same numbers. Also with train: 12:53:39 | loaded 2303 episodes with a total of 60589 examples vs `parlai dd -t fb:google_sgd_rl_splits:OutDomainSystemTeacher' train: 12:22:41 | loaded 2303 episodes with a total of 60589 examples |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Code for processing Google SGD into In-domain and Out-domain data via
build.py, using via agents.Datasets added in this substack:
Test plan:
Regression test,
parlai ddof dataset