Commit Graph

214 Commits

Author SHA1 Message Date
Arec
a07050a748 fixing rendering system and main script to align with new repo changes 2021-11-09 07:15:07 -08:00
Arec
5799d095d9 adding tools to render directly from a policy, updating data generator, adding scratch files 2021-11-09 06:28:42 -08:00
ebuehrle
4c8fb77a91 Larger batches 2021-11-06 18:49:00 +01:00
ebuehrle
bc774c54ca Refactor LL buffer 2021-11-06 16:43:45 +01:00
ebuehrle
b634a34461 RMSprop + weight decay, no discount, action noise 2021-11-05 15:56:33 +01:00
ebuehrle
4a69322ed0 Move eps to constructor 2021-11-05 10:32:07 +01:00
ebuehrle
4ee960b104 Fix rendering 2021-11-05 08:35:14 +01:00
ebuehrle
071c731921 More expert data 2021-11-04 19:54:40 +01:00
ebuehrle
081fb4e6ab Sample with replacement 2021-11-04 19:54:14 +01:00
ebuehrle
f794a96112 Sample expert data from same distribution 2021-11-04 18:52:08 +01:00
ebuehrle
9a95134530 Remove rasterized speed 2021-11-04 17:21:10 +01:00
ebuehrle
0990eb808a Larger batches 2021-11-04 11:01:06 +01:00
ebuehrle
361c1ea9d3 Add speed, then accumulate, skip 5 frames 2021-11-03 17:49:59 +01:00
ebuehrle
2718f52941 Improve custom env support 2021-11-03 17:48:27 +01:00
ebuehrle
ba850eadbb Move rasterized speed env to scratch 2021-11-03 08:48:19 +01:00
ebuehrle
06fb1afae4 Random skip, longer horizon 2021-11-02 16:48:31 +01:00
ebuehrle
2aaaad36f0 Fix rendering
- support predict()
- move discount to OptionsEnv
- fix RenderOptions
2021-11-02 16:15:34 +01:00
ebuehrle
6416fceb60 Fix horizon 2021-11-02 12:57:40 +01:00
ebuehrle
a1ae103ab4 Merge branch 'options-env' into dev 2021-11-02 10:18:33 +01:00
ebuehrle
66c10f5280 Make options env compatible with PPO 2021-11-02 09:33:16 +01:00
ebuehrle
e37447caff Increase discriminator batch size and update steps 2021-10-29 16:33:06 +02:00
ebuehrle
2e8c7f1ace Use rasterized speed 2021-10-29 16:09:03 +02:00
ebuehrle
fc2cd936a8 Copy options to scratch 2021-10-29 14:56:33 +02:00
ebuehrle
8dd42abbf3 use discriminator preprocessing 2021-10-29 14:31:10 +02:00
ebuehrle
92981ba284 Change batch size and discriminator updates
Training not successful
2021-10-29 11:54:32 +02:00
Johannes Fischer
fb7e841dc3 Merge branch 'main' of github.com:sisl/InteractionImitation 2021-10-29 11:35:08 +02:00
ebuehrle
4317cee272 Check emergency braking in available actions 2021-10-29 11:31:38 +02:00
ebuehrle
cedb95dd57 Vectorize action propagation 2021-10-29 11:17:16 +02:00
ebuehrle
e57da1338a Check emergency braking in available actions 2021-10-29 11:10:42 +02:00
ebuehrle
999cb80e13 Merge branch 'vectorize-propagation2' 2021-10-29 11:07:14 +02:00
ebuehrle
5bdf85562b Merge branch 'main' into dev 2021-10-29 10:54:21 +02:00
Johannes Fischer
673b565e11 Update scratch 2021-10-28 18:31:52 +02:00
Johannes Fischer
1a1f6d8836 Merge branch 'main' of github.com:sisl/InteractionImitation 2021-10-28 18:17:47 +02:00
Johannes Fischer
e36aaef9f2 Move and add scratch scripts 2021-10-28 18:16:17 +02:00
Johannes Fischer
0077c24074 precompute expert features 2021-10-28 18:16:00 +02:00
ebuehrle
b1740764e3 Parameterize number of discriminator updates per epoch 2021-10-28 17:58:42 +02:00
ebuehrle
8d7409c914 Check emergency braking in available actions 2021-10-28 17:54:58 +02:00
ebuehrle
f9e058a7d9 Vectorize available actions computation 2021-10-28 15:47:39 +02:00
Johannes Fischer
070b8fc785 Cleanup 2021-10-28 13:49:14 +02:00
Johannes Fischer
c5b043c49f Cleanup 2021-10-28 13:47:54 +02:00
Johannes Fischer
8703b11dee import evaluation 2021-10-28 13:31:30 +02:00
Johannes Fischer
1bab1aaab7 Cleanup evaluation 2021-10-28 13:31:08 +02:00
Johannes Fischer
7ffcc0b4b8 Extract evaluation code to separate file 2021-10-28 13:29:00 +02:00
Johannes Fischer
2218d14409 Move metrics 2021-10-28 13:26:17 +02:00
Johannes Fischer
f217daf251 update evaluation 2021-10-28 13:10:44 +02:00
Johannes Fischer
09afee4e1d Add first metrics 2021-10-28 09:55:20 +02:00
Johannes Fischer
214c6daeaa Merge branch 'main' of github.com:sisl/InteractionImitation 2021-10-28 09:45:27 +02:00
Johannes Fischer
d89c77409c Add some scratch 2021-10-28 09:43:55 +02:00
Johannes Fischer
3857716cec Rename predict to forward
This is done to be consistent with stable baselines interface. predict is then automatically defined. This is necessary to use stable baselines' evaluate_policy method
2021-10-28 09:42:27 +02:00
Johannes Fischer
71f69c43ed Fix typo in dataset folder name 2021-10-28 09:40:52 +02:00