Commit Graph
73 Commits
Author SHA1 Message Date
dodox ecf0b72e88 Port soem fixes / additions learned from implementign itpal_jax 2025-01-22 14:03:23 +01:00
dodox 3816adef9a streamline tests, mitigate broken env binding 2025-01-22 13:46:16 +01:00
dodox 04be117a95 Various fixes 2025-01-22 13:46:00 +01:00
dodox fe7c7b3db0 better handling of missing cpp_projection 2025-01-22 13:45:34 +01:00
dodox 90666a695c add .vscode to gitignore 2025-01-22 13:45:05 +01:00
dodox c1189351cf ensure compat version of gymnasium 2025-01-22 13:44:43 +01:00
dodox e938018494 trl spec for .reset 2024-11-07 11:41:01 +01:00
dodox 4f8fc500b7 Simplify operations on spaces (is_discrete, shape) 2024-11-07 11:40:32 +01:00
dodox 5c44448e53 Use trl space definitions (not gym) 2024-11-07 11:39:45 +01:00
dodox 8a078fb59e Fix: Issue with env wrapping (ensure batch dim) 2024-11-07 11:39:09 +01:00
dodox 52b3f3b71e Updated README 2024-11-07 11:38:34 +01:00
dodox df1ba6fe53 Updated README 2024-10-21 15:25:01 +02:00
dodox 8eb9b384c7 Pytest as optional dep 2024-10-21 15:24:45 +02:00
dodox abc8dcbda1 Expand Tests 2024-10-21 15:24:36 +02:00
dodox e927afcc30 Extend policy impl 2024-10-21 15:24:20 +02:00
dodox ca1ee980ef Fix: Bug in loss calc for TRPLLoss 2024-10-21 15:23:57 +02:00
dodox 0c6e58634f Rework algo impls 2024-10-21 15:23:39 +02:00
dodox 651ef1522f Fixing issues with projections 2024-10-21 15:23:17 +02:00
dodox 71cb8593d9 Fix: Tried to reference gym space classes even if no gym avaible 2024-08-30 08:05:41 +02:00
dodox 906240e145 Fix README typo 2024-08-28 12:15:51 +02:00
dodox af444d85e7 Updated README 2024-08-28 12:15:20 +02:00
dodox e6d78083aa Fix trpl test using wrong hps 2024-08-28 11:58:58 +02:00
dodox 54bab221ef Disable vlearn for now... 2024-08-28 11:55:43 +02:00
dodox 1a02568f3c Rename frob_projection file 2024-08-28 11:55:30 +02:00
dodox 0464fbabe8 Disable vlearn test for now 2024-08-28 11:55:04 +02:00
dodox f7d171399f Updated pyproject.toml 2024-08-28 11:46:31 +02:00
dodox ebff2551cd projections_legacy no longer requierd 2024-08-28 11:34:26 +02:00
dodox 9c55b6a110 remove old vleasrn objective impl 2024-08-28 11:33:30 +02:00
dodox 4f58ce0ff2 refactor algo impls 2024-08-28 11:33:20 +02:00
dodox dd98af9f77 Refactor objective losses 2024-08-28 11:32:51 +02:00
dodox 25988bab54 Added tmp (untested) vlearn objective impl 2024-08-28 11:32:22 +02:00
dodox cb48badcff Adapt init 2024-08-28 11:32:07 +02:00
dodox 5fc4b30ea8 New projection impls 2024-08-28 11:31:42 +02:00
dodox d29417187f Implemenetd new test suite 2024-08-28 11:30:37 +02:00
dodox 416c2036a5 vlearn loss draft 2024-07-17 14:53:11 +02:00
dodox e106d8701f Vlearn draft 2024-07-17 14:52:44 +02:00
dodox 5f279beccf New proj implementations 2024-07-17 14:51:59 +02:00
dodox 4240f611ac Updated TODO 2024-06-02 16:37:15 +02:00
dodox 8d5d44e992 Refactor out some func into general Algo class 2024-06-02 16:36:59 +02:00
dodox 5f186af9fb Fix class name for TRPL 2024-06-02 16:25:18 +02:00
dodox 6cb320f432 TODO and WIP warning for README 2024-06-02 16:18:11 +02:00
dodox 7861821d0d Worked on TRPL module 2024-06-02 14:14:12 +02:00
dodox 65c6a950aa Use loggers correclty 2024-06-02 14:13:36 +02:00
dodox 4091df45f5 Implement rudimentary TerminalLogger 2024-06-02 14:12:41 +02:00
dodox c7f5fcbf0f I hate debugging tensordict weirdness 2024-06-02 13:56:54 +02:00
dodox a867a74138 Reorder... 2024-06-02 12:09:26 +02:00
dodox d51bf948d4 Make sure we work without original gym installed 2024-06-02 12:09:13 +02:00
dodox b4f89c9b7a Added TODO fro module docs 2024-06-02 12:00:19 +02:00
dodox dd6c6b6165 Why was that h3? 2024-06-02 11:59:26 +02:00
dodox 78d79cf705 Initial code fro projections 2024-06-02 11:57:19 +02:00
dodox add8e92b4a Add norms 2024-06-02 11:52:20 +02:00
dodox c6a12aa27b Started TRPL impl 2024-06-02 11:49:09 +02:00
dodox 3931f5e31b Started implementation of TRPL loss objective module 2024-06-02 11:48:51 +02:00
dodox b34224f189 refactor dir structure 2024-06-02 11:13:04 +02:00
dodox 8a013a4746 Expanded .gitignore 2024-06-02 11:11:35 +02:00
dodox 0a95ea652a Support for torchrl envs 2024-06-02 11:11:18 +02:00
dodox 50733bb1a4 Make use of new utils to work with any space 2024-06-02 11:07:57 +02:00
dodox 59060c7533 refactor 2024-06-02 11:07:46 +02:00
dodox a3cca71ac9 Add utils to handle all known space types 2024-06-02 11:06:55 +02:00
dodox 5a6069daf4 Showcase TRPL in README 2024-05-31 18:25:42 +02:00
dodox 1086c9f6fd Remove all etsts for now (interface changed) 2024-05-31 18:25:17 +02:00
dodox 015f1e256a Refactor 2024-05-31 18:25:03 +02:00
dodox 0bf748869a Switch to pyproject.toml 2024-05-31 18:24:47 +02:00
dodox bd507c37c3 native support for fancy_gym 2024-05-31 13:11:09 +02:00
dodox 2e7859f068 Fixed README 2024-05-31 13:06:50 +02:00
dodox 360d2569f0 Updated README 2024-05-31 13:05:11 +02:00
dodox 0808655136 Update example.py to new func 2024-05-31 13:04:56 +02:00
dodox 1d8d217ec0 Refactor env handling 2024-05-31 13:04:41 +02:00
dodox 1d1d9060f9 Switch to using torchrl loggers 2024-05-31 13:04:26 +02:00
dodox 7ea0bdcec6 Remove utils.py 2024-05-31 13:04:02 +02:00
dodox dc70d045ab Updated README 2024-05-30 22:26:36 +02:00
dodox 3082d1bed5 Icon? 2024-05-29 21:41:36 +02:00
dodox 8946362336 Oh, I could start using git... 2024-05-29 21:21:43 +02:00