Commit Graph
14 Commits
Author SHA1 Message Date
Kittipat Virochsiri fdc48c9071 AutoDataModule
Reviewed By: czxttkl

Differential Revision: D24463611

fbshipit-source-id: e4989c78ac5e2979b715bc53cb46707e157bc0ca
2020-12-01 15:38:18 -08:00
John Reese 2163bd7d0a apply black 20.8b1 formatting update
Summary:
allow-large-files

black_any_style

Reviewed By: zertosh

Differential Revision: D24325133

fbshipit-source-id: b4afe80d1e8b2bc993f4b8e3822c02964df47462
2020-10-14 20:22:08 -07:00
Kaiwen Wang b3775bc8ec Continuous VM Tuning (#298)
Summary:
Pull Request resolved: https://github.com/facebookresearch/ReAgent/pull/298

A continuous actor takes in user + N candidate features for PVM

Reviewed By: czxttkl

Differential Revision: D22428244

fbshipit-source-id: e9f8dde9ce16489ee1d0b52cd3b15751077d32f9
2020-10-02 14:27:07 -07:00
Zhengxing Chen 0d294b11e5 Back out recent refactor
Summary:
Need more tests before landing the refactor diffs: D22702504 (https://github.com/facebookresearch/ReAgent/commit/1b470c489d19c33beab88b8ea2e79843d4d31f28), D23123762 (https://github.com/facebookresearch/ReAgent/commit/76829287265bc39f879f3bc1d946a1374c5e1141), D23124179 (https://github.com/facebookresearch/ReAgent/commit/b28f84aa013be00194508f52498160592cb37e9d), D23219012 (https://github.com/facebookresearch/ReAgent/commit/e404c5772ea4118105c2eb136ca96ad5ca8e01db)

Back out to a version based on D23155753.

Check our team diff history: https://fburl.com/diffs/ppsgazgj

Reviewed By: kittipatv

Differential Revision: D23270626

fbshipit-source-id: 14653066bb3924a987a54650a51241895b321c8e
2020-08-21 15:59:42 -07:00
jjg e404c5772e Merge types.py into core/types.py. Consolidate registration into ml/rl/register.py
Reviewed By: czxttkl

Differential Revision: D23219012

fbshipit-source-id: 0bc65ee8a017145c0a655953b19d203c173996c2
2020-08-20 00:21:37 -07:00
Kaiwen Wang f4e372ad14 DQN Sparse Features pt3: Offline workflow
Summary: Offline workflow for Sparse DQN

Reviewed By: czxttkl

Differential Revision: D21966074

fbshipit-source-id: 29c3e6a7b5b03465c0d5f396b3bfb9cb069bd893
2020-07-03 21:44:38 -07:00
Zhengxing Chen 96073b457c Seq2Slate Distributed Batch Preprocessor
Summary: This is the first step to use a distributed data reader: build a torchscript batch preprocessor.

Reviewed By: kaiwenw

Differential Revision: D22036324

fbshipit-source-id: d45980e9cb2863579a1b10d7cfd8c3f2cb610ca5
2020-06-15 22:38:20 -07:00
Kaiwen Wang 43336869f0 Implement WorldModel preprocessor with transforms
Summary: Also code cleanup, such as removing RawFeatures

Reviewed By: czxttkl

Differential Revision: D21946302

fbshipit-source-id: 755037338f8cab85b049967727bc652e4cdcf3f2
2020-06-10 16:09:46 -07:00
Kaiwen Wang c5de55e7b9 Update DiscreteDQNBatchPreprocessor to use transforms
Reviewed By: czxttkl

Differential Revision: D21920081

fbshipit-source-id: 6c6f109898d5443cd1a420983cef153da7384b9b
2020-06-10 16:09:25 -07:00
Kaiwen Wang af7e0d65d4 Delete part of old gym and reorganize ModelManager (#266)
Summary:
Pull Request resolved: https://github.com/facebookresearch/ReAgent/pull/266

- replace normalization parameters with normalization data
- clean up _set_normalization_data
- clean up train_workflow
- deprecate OpenAIGym, OpenAIMemoryPool

Reviewed By: kittipatv

Differential Revision: D21578576

fbshipit-source-id: 32c780e980b915ab0bea70348bd79f98c48fd669
2020-05-20 10:01:33 -07:00
Kaiwen Wang 1748e87d0f Implement and test PDQN for OSS (#264)
Summary:
Pull Request resolved: https://github.com/facebookresearch/ReAgent/pull/264

- Move PDQN from fblearner folder to OSS. Fill in ModelManager for it.
- Add Gym tests for maxq and sarsa
- Clean up ParametricBatchPreprocessor to use transforms
- Delete RawTrainingBatch
- Delete open_ai_gym_tests :)

Reviewed By: kittipatv

Differential Revision: D21481434

fbshipit-source-id: b895d3f20d72fb0aa22163c5220654c8b5bfd21b
2020-05-19 21:26:25 -07:00
Kai Wen Wang f2e39855d7 Move SAC Model Manager for online gym tests. (#241)
Summary: Pull Request resolved: https://github.com/facebookresearch/ReAgent/pull/241

Reviewed By: kittipatv

Differential Revision: D21248464

Pulled By: kaiwenw

fbshipit-source-id: b5664df1b5af55036551d44877914fca026a30b2
2020-05-02 08:04:53 -07:00
Kittipat Virochsiri 811e1210ff Use transforms Seq2Slate batch preprocessor
Summary:
- Removed manual squeezing in seq2slate code path; ensure that the 0th dim of tensor corresponds to batch size
- Use transforms in Seq2Slate batch preprocessor

Reviewed By: czxttkl

Differential Revision: D21224914

fbshipit-source-id: 9935d1641d0f9109e998775385cac821bc8c5a06
2020-04-28 07:36:21 -07:00
Kittipat Virochsiri de10f79e39 Write SlateQ batch preprocessor as transforms
Summary: Make it easier to compose transformation

Reviewed By: badrinarayan

Differential Revision: D21198737

fbshipit-source-id: 158b956bc7148e27aeb36108257bccde7502e67f
2020-04-23 16:01:53 -07:00