fe4472cb
Add decoders, distributions, encoders, layers, networks, and utils
4 年前
89890bf2
Update with newest changes
4 年前
e3bc3352
[pytorch] Add decoders, distributions, encoders, layers, networks, and utils ( #4349 )
4 年前
1955af9e
[feature] Add experimental PyTorch support ( #4335 )
* Begin porting work
* Add ResNet and distributions
* Dynamically construct actor and critic
* Initial optimizer port
* Refactoring policy and optimizer
* Resolving a few bugs
* Share more code between tf and torch policies
* Slightly closer to running model
* Training runs, but doesn’t actually work
* Fix a couple additional bugs
* Add conditional sigma for distribution
* Fix normalization
* Support discrete actions as well
* Continuous and discrete now train
* Mulkti-discrete now working
* Visual observations now train as well
* GRU in-progress and dynamic cnns
* Fix for memories
* Remove unused arg
* Combine actor and critic classes. Initial export.
* Support tf and pytorch alongside one another
* Prepare model for onnx export
* Use LSTM and fix a few merge errors
* Fix bug in probs calculation
* Optimize np -> tensor operations
* Time action sample funct...
4 年前
9f51ab14
Saving the reward providers
4 年前
25454a48
adding tests
4 年前
108fac9a
Replace torch.detach().cpu().numpy() with a utils method
4 年前
328353bc
Torch : Saving/Loading of the reward providers ( #4405 )
* Saving the reward providers
* adding tests
* Moved the tests around
* Update ml-agents/mlagents/trainers/tests/torch/saver/test_saver_reward_providers.py
* Update ml-agents/mlagents/trainers/tests/torch/saver/test_saver_reward_providers.py
* Update ml-agents/mlagents/trainers/tests/torch/saver/test_saver_reward_providers.py
Co-authored-by: Ruo-Ping (Rachel) Dong <ruoping.dong@unity3d.com>
* Update ml-agents/mlagents/trainers/tests/torch/saver/test_saver_reward_providers.py
Co-authored-by: Ruo-Ping (Rachel) Dong <ruoping.dong@unity3d.com>
Co-authored-by: Ruo-Ping (Rachel) Dong <ruoping.dong@unity3d.com>
4 年前
31750e97
Using item() in place of to_numpy()
4 年前
6587c911
hybrid bheavior spec
4 年前
88eff042
Merge branch 'master' into develop-saver-name
4 年前
f4f1a8d9
merge master into trainer-plugin branch
4 年前
12e15e29
Fix on GAIL Torch when using actions ( #4407 )
4 年前
498934f9
Replace torch.detach().cpu().numpy() with a utils method ( #4406 )
* Replace torch.detach().cpu().numpy() with a utils method
* Using item() in place of to_numpy()
* more use of item() and additional tests
4 年前
5b1620e9
moved encoders to processors
4 年前
27fb4270
brain_name to behavior_name
4 年前
1dca75d8
Move linear encoding to NetworkBody
4 年前
bfda9576
Replace brain_name with behavior_name ( #4419 )
brain_name -> behavior_name
some prob -> log_prob in comments
rename files optimizer -> optimizer_tf for tensorflow
4 年前
9364048e
Merge branch 'develop-pytorch-encoding' into develop-restructure-torch-networks
4 年前
fd1dc3a6
Merge branch 'master' into develop-torch-omp
4 年前
f5dee9d1
jit for continuous control
4 年前
ef3be79e
sac
4 年前
7b4d0865
[Bug fix] Fix bug in GAIL gradient penalty ( #4425 )
4 年前
4e93cb6e
[torch] Restructure PyTorch encoders ( #4421 )
* Move linear encoding to NetworkBody
* moved encoders to processors (#4420 )
* fix bad merge
* Get it running
* Replace mentions of visual_encoders
* Remove output_size property
* Fix tests
* Fix some references
* Revert test_simple_rl
* Fix networks test
* Make curiosity test more accomodating
* Rename total_input_size
* [Bug fix] Fix bug in GAIL gradient penalty (#4425 ) (#4426 )
Co-authored-by: Vincent-Pierre BERGES <vincentpierre@unity3d.com>
* Up number of steps
* Rename to visual_processors and vector_processors
Co-authored-by: andrewcoh <54679309+andrewcoh@users.noreply.github.com>
Co-authored-by: Andrew Cohen <andrew.cohen@unity3d.com>
Co-authored-by: Vincent-Pierre BERGES <vincentpierre@unity3d.com>
4 年前
beb5eb30
[bug-fix] Fixes for Torch SAC and tests ( #4408 )
* Fixes for Torch SAC and tests
* FIx recurrent sac test
* Properly update normalization for SAC-continuous
* Fix issue with log ent coef reporting in SAC Torch
4 年前
6f534366
Add torch_utils class, auto-detect CUDA availability ( #4403 )
* Add torch_utils
* Use torch from torch_utils
* Add torch to banned modules in CI
* Better import error handling
* Fix flake8 errors
* Address comments
* Move networks to GPU if enabled
* Switch to torch_utils
* More flake8 problems
* Move reward providers to GPU/CPU
* Remove anothere set default tensor
* Fix banned import in test
4 年前
676f5f7c
[refactor] Refactor GAIL to use new encoder structure ( #4433 )
Co-authored-by: Ervin Teng <ervin@unity3d.com>
Co-authored-by: Vincent-Pierre BERGES <vincentpierre@unity3d.com>
4 年前
60eacc0d
Merge branch 'master' into develop-adjust-cpu-settings
4 年前
6986fb10
use LinearEncoder in curiosity and clean up ( #4444 )
4 年前
fdc887a1
Some experimental stuff
4 年前
bf6506fc
[feature] Add small CNN for grids 5x5 and up ( #4434 )
4 年前
3997b14b
Merge branch 'master' into develop-hybrid-actions
4 年前
069f10b9
[Bug fix] Export all branches for discrete control torch ( #4491 )
* Export all branches for discrete control torch
* [skip ci] Changelog edits
* Update ml-agents/mlagents/trainers/torch/networks.py
Co-authored-by: Ruo-Ping (Rachel) Dong <ruoping.dong@unity3d.com>
* Update ml-agents/mlagents/trainers/torch/networks.py
* Fix formatting
Co-authored-by: Ruo-Ping (Rachel) Dong <ruoping.dong@unity3d.com>
4 年前
14a7e29b
Add AMRL layer
4 年前
8294bc59
ignore commit
4 年前
fc557903
Develop fix visual torchh export ( #4494 )
* Fixing exporting of ONNX for visual when using threading
* docstring was wrong
4 年前
3e771cbb
Permute visual obs outside of network
4 年前
9f96a495
Use built-in cumulative max
4 年前
b0b2e22e
prepare to merge action_out
4 年前
77c810fb
Fix SAC and make utility method
4 年前
43c41d66
Fix BC and Reward Signals
4 年前
7754ad7b
Don't run value during inference
4 年前
3f7d815c
detach masks
4 年前
10f3e1c7
Cherry picked #4491 ( #4493 )
[Bug fix] Export all branches for discrete control torch
4 年前
37645aa2
Develop fix visual torchh export ( #4494 ) ( #4497 )
* Fixing exporting of ONNX for visual when using threading
* docstring was wrong
4 年前
85602279
add action_out to dist
4 年前
b6095151
Execute critic with LSTM
4 年前
181bdec0
-
4 年前
4e4ad7b0
Don't run value during policy evaluate, optimized soft update function ( #4501 )
* Don't run value during inference
* Execute critic with LSTM
* Address comments
* Unformat
* Optimized soft update
* Move soft update to model utils
* Add test for soft update
4 年前
f0a2c6f2
add action_out abstract method and docstring
4 年前
f9ff3efe
Merge branch 'develop-policyonly' into develop-sac-targetq
4 年前
7c0aa77b
Merge branch 'develop-actions-out' into develop-hybrid-actions
4 年前
60b76790
Random Network Distillation for Torch ( #4473 )
* initial commit
* works with Pyramids
* added unit tests and a separate config file
* Adding first batch of documentation
* adding in the docs that rnd is only for PyTorch
* adding newline at the end of the config files
* adding some docs
* Code comments
* no normalization of the reward
* Fixing the tests
* [skip ci]
* [skip ci] Make sure RND will only work for Torch by editing the config file
* [skip ci] Additional information in the Documentation
* Remove the _has_updated_once flag
4 年前
b6b453a0
rename to exported model output
4 年前
9cbf6c3f
Merge branch 'develop-actions-out' into develop-hybrid-actions
4 年前
72cd0d39
separate tensors for disc/cont
4 年前
643c8e58
ppo extended
4 年前
ba29b6b4
Change dimensions of recurrent in to BxN for ONNX
4 年前
400e14cb
[Bug-fix] RND would not be saved correctly. Added tests ( #4514 )
4 年前
db37db34
fixing errors
4 年前
44c9879e
action models
4 年前
a3bf96fd
Merge branch 'master' into hh/develop/gridsensor-tests
4 年前
c494bfcc
trains successfully
4 年前
e8431a6d
Proper dimensions for entropy, sum before bonus in PPO
4 年前
190d8e4d
action model as a singleton
4 年前
8dec4771
Add hybrid actions to SAC
4 年前
d6544353
using lists for distributions
4 年前
e686a785
removed abstract class
4 年前
81342148
Revert "Add hybrid actions to SAC"
This reverts commit a759b36a51df4f8f1fd296f9f148269f0f026e42.
4 年前
74553853
use ModuleList
4 年前
aa5cf8ab
rename ActionModel
4 年前
4b9a7db6
remove old behaviorspec
4 年前
5a3f5b4a
remove comments
4 年前
e5f14400
Merge branch 'master' into develop-hybrid-actions-singleton
4 年前
e4db5dc5
ActionSpec and ActionBuffer ( #4578 )
4 年前
eaecb59e
torch utils to and from buffer
4 年前
6e23bafd
ActionFlattener Refactor
4 年前
8013e544
ignoring Instance of 'AbstractContextManager' has no 'enter_context' member (no-member)
4 年前
e0ef30a5
[bug-fix] Change entropy computation and loss reporting in Torch to match TF ( #4538 )
* Proper dimensions for entropy, sum before bonus in PPO
* Make entropy reporting same as TF
* Always use separate critic
* Revert to shared
* Remove unneeded extra line
* Change entropy shape in test
* Change another entropy shape
* Add entropy summing to evaluate_actions
* Add notes about torch.abs(policy_loss)
4 年前
cb8e4d25
Add ActionSpec ( #4586 )
Co-authored-by: Ervin T <ervin@unity3d.com>
4 年前
b40e7793
fix mlagents-envs tests
4 年前
a4ba90ec
[bug-fix] Fix ONNX export/Barracuda import for continuous actions ( #4608 )
* Use torch.ones rather than expand
* use tf.cat instead of tf.expand for Barracuda
* fix error msg handling
* handle loading exception
Co-authored-by: Chris Elion <chris.elion@unity3d.com>
4 年前
9689cf2c
remove *_action_* from function names
4 年前
64e998a2
[bug-fix] Use float64 when converting np.ndarray to torch.tensor, cap Torch version to 1.7.x ( #4610 )
* Use float64 in GAIL tests
* Use float32 when converting np arrays by default
* Enforce torch 1.7.x or below
* Add comment about Windows install
* Adjust tests
4 年前
96452986
Initial commit for multi head attention
4 年前
dc89318d
remove ActionType
4 年前
a3a9a56b
Merge branch 'exp-multi-head-attention' into exp-bullet-hell
4 年前
9e08be87
Merge branch 'master' into release_9_branch_merge
4 年前
99ec16e6
Hard Swish
4 年前
6cf54bf2
remove self.action_spec from policy/bc
4 年前
d3d4eb90
Trainer with attention
4 年前
5d3ad161
Leaky ReLU
4 年前
97dfa142
fix action_spec refs
4 年前
7ef3c9a1
Trainer with attention
4 年前
f00ff21d
Add clipping to Torch
4 年前
b853e5ba
Action buffer ( #4612 )
Co-authored-by: Ervin T <ervin@unity3d.com>
Co-authored-by: Vincent-Pierre BERGES <vincentpierre@unity3d.com>
4 年前
b09dfd0a
Still not working
4 年前
7e6cc222
Take mean of continuous entropy
4 年前
e80d418b
Use lower scaling value
4 年前
89489ae0
Invert divide by 3 in log prob
4 年前
3c96a3a2
Action Model ( #4580 )
Co-authored-by: Ervin T <ervin@unity3d.com>
Co-authored-by: Vincent-Pierre BERGES <vincentpierre@unity3d.com>
4 年前
0e28dd8f
add static method to create continuous/discrete
4 年前
c264b4da
additional changes
4 年前
7502a383
Proper dimensions
4 年前
c3cec801
Use linear gain for KaimingHe
4 年前
1f305f24
Test commit
4 年前
88d3ec3e
Merge master into hybrid actions staging branch ( #4704 )
4 年前
48a11a0e
Undo test commit
4 年前
66f3d2ad
Merge branch 'develop-contentropy' into develop-torch-clip
4 年前
ae920478
resolve conflicts
4 年前
9b3d4ade
Fixing some issues
4 年前
87a7ccf8
use int64 steps, check for NaN actions ( #4607 )
* use int64 steps
* check for NaN actions
Co-authored-by: Ruo-Ping Dong <ruoping.dong@unity3d.com>
4 年前
db6ebff6
Remove clip
4 年前
23800f33
Merge branch 'master' into develop-action-spec
4 年前
85a7c0f7
[bug-fix] Add clipping to PyTorch policy, fix initialization ( #4649 )
4 年前
08c8862e
Add epsilon to log
4 年前
0548057d
Use real clipping (as in TF)
4 年前
0cdb2040
Use tanh squash
4 年前
719d94e9
fix entropy_sum after merge
4 年前
6fcbba53
Refactoring the code to make it more flexible. Still a hack
4 年前
eb4f3065
Add another epsilon
4 年前
733bffbf
use int64 steps, check for NaN actions ( #4607 ) ( #4654 )
* use int64 steps
* check for NaN actions
Co-authored-by: Ruo-Ping Dong <ruoping.dong@unity3d.com>
Co-authored-by: Chris Elion <chris.elion@unity3d.com>
4 年前
23b42dcd
Merge branch 'develop-torchcrawlerdebug' into develop-torch-clip
4 年前
58f38662
Bigger network, needed to solve
4 年前
efdb1042
Add global lock for torch.onnx.export() ( #4659 )
4 年前
adc039a5
Use linear gain for KaimingHe
4 年前
0b6c2ed3
Fixing some bugs
4 年前
8175d558
[bug-fix] Fix BC module + action clipping ( #4667 )
4 年前
2a8c6800
[bug-fix] Add clipping to PyTorch policy, fix initialization ( #4649 ) ( #4662 )
4 年前
3b3b53e2
Improve comment
4 年前
c74adb81
fix specs in torch util
4 年前
84cc2b84
concat x self before attention
4 年前
e14e1c4d
Improvements and new tests
4 年前
de27d7a6
[bug-fix] Fix issue where NaNs are outputted by the policy when training Match3 ( #4664 )
* match3 settings
* Add epsilon to log
* Add another epsilon
* Revert match3 configs
* NaN-free masking method
* Add comment for paper
* Add comment for paper
Co-authored-by: Chris Elion <chris.elion@unity3d.com>
4 年前
953cb6bb
Merge branch 'master' into develop-windows-delay
4 年前
8df04e83
Add global lock for torch.onnx.export() ( #4665 )
* Cherry-pick fix from #4659
4 年前
d62f6b0a
modify bullet/attn
4 年前
1cac78e2
[Bug-fix] Remove extra directory created at checkpoint ( #4675 )
* [Bug-fix] Remove extra directory created at checkpoint
* removing os import
4 年前
bc746839
Normalize GAIL observations
4 年前
ee5313e4
Merge branch 'master' into develop-windows-delay
4 年前
78f88c15
Add clip to export and make optional in policy
4 年前
f283cb60
different architecture
4 年前
362f2ec0
Use correct dimensions of gradient
4 年前
f0ed3a38
Cherry-pick BC fixes to Release 10 ( #4668 )
4 年前
b6c903c8
Cherry-pick NaN fix for Match 3 ( #4664 )
* match3 settings
* Add epsilon to log
* Add another epsilon
* Revert match3 configs
* NaN-free masking method
* Add comment for paper
* Add comment for paper
Co-authored-by: Chris Elion <chris.elion@unity3d.com>
Co-authored-by: Chris Elion <chris.elion@unity3d.com>
4 年前
3eba7423
Increase initialization
4 年前
fd31da37
[Bug-fix] Remove extra directory created at checkpoint ( #4675 ) ( #4678 )
* [Bug-fix] Remove extra directory created at checkpoint
* removing os import
4 年前
4f8f21e4
Bigger scale
4 年前
3f771e61
add ActionBuffers and utils
4 年前
7a0ebfbd
Pretty broken
4 年前
95bdbba3
Less broken PPO
4 年前
6c77ac7a
Update SAC, fix PPO batching
4 年前
278911a5
Fix staging tests ( #4708 )
4 年前
1db21cbb
Fix SAC interrupted condition and typing
4 年前
94c59e31
C# changes for hybrid action spaces ( #4587 )
* Add hybrid action capability flag (#4576 )
* Change BrainParametersProto to support ActionSpec (#4579 )
* Assign new BrainParametersProto fields based on capabilities (#4581 )
* ActionBuffer with hybrid actions for RemotePolicy (#4592 )
* Barracuda inference for hybrid actions (#4611 )
* Refactor BarracudaModel loader checks (#4629 )
* Export separate nodes for continuous/discrete actions (#4655 )
* Separate continuous/discrete actions in AgentActionProto (#4698 )
* Force different nodes for new and deprecated action output (#4705 )
4 年前
4158629e
Properly feed in None rather than empty arrays
4 年前
bd917c9c
action buffer passes continuous
4 年前
90da7426
Fix masking for torch
4 年前
5a5bd515
Fix multiple obs
4 年前
b36fcf16
discrete runs/cont passes
4 年前
e85d8e35
Added a comment and included the change of #4715 for simplicity
4 年前
8d29114d
Update curiosity reward provider
4 年前
ad951493
debugging discrete
4 年前
79a3051e
Update GAIL and BC
4 年前
fcf6471e
2d discrete passes
4 年前
2c744129
Merge pull request #4714 from Unity-Technologies/develop-fix-mask
Fix masking for torch
4 年前
fdaa8c3d
Merge branch 'develop-unified-obs' into develop-centralizedcritic
4 年前
056630d7
sac continuous and discrete train
4 年前
990f801a
Develop hybrid action staging ( #4702 )
Co-authored-by: Ervin T <ervin@unity3d.com>
Co-authored-by: Vincent-Pierre BERGES <vincentpierre@unity3d.com>
Co-authored-by: Ruo-Ping Dong <ruoping.dong@unity3d.com>
Co-authored-by: Chris Elion <chris.elion@unity3d.com>
4 年前
735fcd52
[WIP] Refactor trainers to use list of obs rather than vec and vis obs
4 年前
6846af21
Multi-input network
4 年前
85e4db33
bc tests pass
4 年前
93ca1409
fixing the tests
4 年前
7a5cc9ec
Merge master into develop-rm-tf
4 年前
cb4b7ed3
Some minor tweaks but still broken
4 年前
24fd9b3c
torch reward providers all pass
4 年前
12619155
added some docstrings
4 年前
dee6b805
fixed bug in discrete
4 年前
c1587bce
Solving merge conflicts
4 年前
2b8ab09d
Use Vince's ONNX export code
4 年前
558bfb70
Cleanup
4 年前
b5d1c071
Merge branch 'master' into develop-action-buffer
4 年前
0d2f8887
Merge remote-tracking branch 'origin/master' into goal-conditioning
# Conflicts:
# ml-agents-envs/mlagents_envs/base_env.py
# ml-agents-envs/mlagents_envs/rpc_utils.py
# ml-agents/mlagents/trainers/tests/mock_brain.py
# ml-agents/mlagents/trainers/tests/simple_test_envs.py
4 年前
73b778cc
rename extract to from_dict
4 年前
cc6b4564
Multi Directional Walker and Initial Hypernetwork ( #4740 )
4 年前
25dfd883
Merge branch 'master' into develop-centralizedcritic
4 年前
ad5f878c
[refactor] Remove critic pass during inference ( #4743 )
4 年前
11687f8d
[cherry-pick] Cherry-pick #4743 into Release 11 ( #4756 )
4 年前
22658a40
use sensor types to differentiate obs ( #4749 )
4 年前
14378aa5
Merging master
4 年前
0c81006d
addressing comments
4 年前
ecac7ca1
Set num teammates to 2
4 年前
95441b75
Use separate hypernetwork module
4 年前
03ee5833
black formatting
4 年前
eef14922
discrete/contionuous unity envs train
4 年前
79898e06
Use hypernetwork in both places
4 年前
f36b10d8
Fix typo
4 年前
a545859e
fix torch test policy
4 年前
43387daa
remove unused import
4 年前
8cb050ef
WIP Made initial changes to enale dimension properties and added attention module
4 年前
6f4f6044
Fix model inference issue with Barracuda v1.2.1 ( #4766 ) ( #4768 )
Co-authored-by: Ervin T. <ervin@unity3d.com>
4 年前
170c269e
Remove prints
4 年前
e168bb4f
add docstrings to AgentAction and ActionLogProbs
4 年前
9fbf56e2
Fix model inference issue with Barracuda v1.2.1 ( #4766 )
Co-authored-by: Ervin T. <ervin@unity3d.com>
4 年前
498b1ee6
Merge branch 'develop-action-buffer' into develop-hybrid-actions-singleton
4 年前
a73f7d73
Turn down gain on GAIL discriminator output ( #4762 )
4 年前
06f1f254
1:1 and continuous/discrete train
4 年前
7750bccd
all hybrid simple rl tests pass
4 年前
2462db69
Add comment to ghost trainer
4 年前
559609a2
Update ml-agents/mlagents/trainers/torch/utils.py
Co-authored-by: Vincent-Pierre BERGES <vincentpierre@unity3d.com>
4 年前
6174c428
move action model to explicit distributions
4 年前
b6bb01b9
Turn down gain on GAIL discriminator output ( #4762 ) ( #4772 )
4 年前
c3699de8
merging master and addressing comments
4 年前
f207a1a6
Revert "Add comment to ghost trainer"
This reverts commit 292b6ce672a7a23ebfdd76bf07b11c50a825dcf4.
4 年前
e0cf88e6
Made a nicer comment
4 年前
f2c9d184
fix set_actions_for_agent
4 年前
29d94c7c
Merge pull request #4734 from Unity-Technologies/develop-obs-as-list
Refactor trainers to use list of obs rather than vec and vis obs
4 年前
e4e0e2bd
Scale size of CC network
4 年前
1d234d1d
bc works
4 年前
719c969c
addressing comments. ObservationSpec is no longer a list
4 年前
a3885120
Scale value network based on num agents
4 年前
8d7e449f
torch curiosity tests pass
4 年前
4bba4e8e
Renaming ObservationSpec to SensorSpec
4 年前
7973b46c
remove print bc
4 年前
c0d01baf
Merge branch 'master' into merge-release11-master
4 年前
95566e44
Merge branch 'develop-action-buffer' into develop-hybrid-actions-singleton
4 年前
c5a057d2
renaming obs_spec variables
4 年前
0cc4eed4
Merge branch 'develop-centralizedcritic' into hh/develop-pushblockcollab
4 年前
e81e68de
comms agent and fixed hallway
4 年前
44ed3258
Merging master
4 年前
ca5a5194
soccer comms on the cloud
4 年前
12828bdc
remove tau from diff for
4 年前
8543b917
add kickpower bonus
4 年前
708ac9bf
use hard=true gbsm
4 年前
449712b0
renaming sensor_spec to sensor_specS
4 年前
35769b53
Merge branch 'develop-action-buffer' into develop-hybrid-actions-singleton
4 年前
c843e3d4
hallway collab exps on cloud
4 年前
272affe0
preliminary aciton model tests
4 年前
a20287f7
continuous comms
4 年前
17496265
move AgentAction, ActionLogProbs, and ActionFlattener to separate files
4 年前
f7a4a31f
[Experiment] Bullet hell
4 年前
76ebc20c
Merge remote-tracking branch 'origin/master' into r12-to-master
4 年前
2dc2ffe3
add action util files
4 年前
f57875e0
layer norm
4 年前
458fee17
Merge pull request #4763 from Unity-Technologies/develop-att
WIP Made initial changes to enable dimension properties and added attention module
4 年前
36cc4665
Removing some vis and vec fields from policy.py
4 年前
bc77c990
layer norm and weight decay with fixed architecture
4 年前
e344fe79
Make memory contiguous ( #4804 )
4 年前
ab0dd150
formatting
4 年前
330fc1d0
Merge branch 'master' into develop-centralizedcritic-mm
4 年前
519c5f47
merging master
4 年前
fad37dc5
add default args to LinearEncoder
4 年前
ad439fb6
Additional changes
4 年前
a482a47c
fixing tests
4 年前
21365c04
formatting
4 年前
d02a1033
Some more fixes
4 年前
8ed14762
Merge branch 'develop-hybrid-actions-singleton' into develop-hybrid-actions-csharp
4 年前
7ba10239
remove action spec attribute from policy
4 年前
96c01a63
custom layer norm
4 年前
efa67290
Self-attention Centralized Critic
4 年前
662fd6b1
added docstrings to action flattener
4 年前
a272bff0
Merge pull request #4803 from Unity-Technologies/develop-remove-vec-vis-fields
Remove some vis and vec fields from Policy
4 年前
d7c3022d
Export separate nodes for continuous/discrete actions ( #4655 )
4 年前
d0c8b5f1
added docstrings to action model
4 年前
90e0e8b2
concat self
4 年前
5c5539af
add zomb scene
4 年前
cc948a41
Policy output actiontuple ( #4651 )
4 年前
89ed777e
increase bullets to 40
4 年前
886883b3
Merge branch 'develop-hybrid-action-staging' into develop-hybrid-actions-singleton
4 年前
60816204
revert changes to 20 bullets
4 年前
231328ea
remove warning prints
4 年前
9689449f
Refactor of attention ( #4840 )
Co-authored-by: Vincent-Pierre BERGES <vincentpierre@unity3d.com>
4 年前
010803ea
separate entity encoder and RSA
4 年前
0b4b0992
Rename more files
4 年前
3ca65063
clean up args in mha
4 年前
7c37c759
Fix some mis-renamings
4 年前
701c1a3f
fix test torch distributions
4 年前
b7e6efa3
Allow setting maximum number of elements in self-attention to None ( #4841 )
* separate entity encoder and RSA
* clean up args in mha
* more cleanups
* fixed tests
* entity embeddings have no max option
* Add exceptions for variable export
* Fix test
* Add docstrings
Co-authored-by: Andrew Cohen <andrew.cohen@unity3d.com>
4 年前
56972f56
WIP integrate attention to networkbody
4 年前
540b930b
add defaults to linear encoder, initialize ent encoders
4 年前
9ae8a720
more cleanups
4 年前
aba633b2
Merge branch 'develop-attention-refactor' into develop-centralizedcritic-mm
4 年前
d27459bd
Fix error in embedding
4 年前
631ac7f4
fixed tests
4 年前
30a09c6f
Merge branch 'develop-attention-refactor' into develop-centralizedcritic-mm
4 年前
eb78a477
Add default init/gain to LinearEncoder ( #4846 )
4 年前
135914fd
entity embeddings work with no max
Integrate into CC
4 年前
0ac990e0
add LayerNorm ( #4847 )
4 年前
af58753b
adding encoding to self and var len input
4 年前
e05e897c
One layer for entity embed
4 年前
8c1a33a4
Fixing .contiguous missing in resnet
4 年前
edbac259
Fixing merge conflicts
4 年前
9a3598eb
Use 4 heads
4 年前
a7d04be6
Merge branch 'develop-hybrid-actions-singleton' into develop-hybrid-actions-csharp
4 年前
d30e9f28
Merge pull request #4851 from Unity-Technologies/fix-contiguous-resnet
Fixing .contiguous missing in resnet
4 年前
5039b65a
Merge branch 'master' into develop-att-network-integration
4 年前
0a876b9c
Fix typos
4 年前
2085e17c
Merge branch 'master' into develop-centralizedcritic-mm
4 年前
d3348074
fix policy action_spec typo
4 年前
e7024786
Removing hacky layer norm
4 年前
e3de0406
Plurals
4 年前
dfe07e1d
Merge branch 'develop-lin-enc-def' into develop-centralizedcritic-mm
4 年前
180d3e20
Merge branch 'develop-centralizedcritic-mm' into develop-cc-teammanager
4 年前
0024a286
merge ervin's new stuff
4 年前
12e1fc28
[feature] Hybrid SAC ( #4574 )
4 年前
b0c02ee0
Merge branch 'develop-hybrid-actions-csharp' into develop-actionmodel-csharp
4 年前
9c3da1b6
New buffer layout, TeamObsUtil, pad dead agents
4 年前
67ad9651
Merge pull request #4825 from Unity-Technologies/sensor-types
[WIP] Observation Types
4 年前
8660b1c2
merging master
4 年前
eab7e42a
Use NaNs to get masks for attention
4 年前
3daa17a9
Merge branch 'develop-centralizedcritic-mm' into develop-zombieteammanager
4 年前
da0c8b9d
Add hypernetwork
4 年前
bf7195f1
Refactor multi input network slightly
4 年前
1c557bd3
Don't concat self
4 年前
38fc2536
addresing some comments
4 年前
6b8b3db3
Try subtract marginalized value
4 年前
92fc78a5
Use new trajectory
4 年前
b2c498de
Use NaNs to get masks for attention
4 年前
e9ff7705
Fix torch tensor non-contiguous issue ( #4855 )
* add contiguous
* preserve tensor device type
4 年前
ad807327
refactor entityembedding/network body
4 年前
01e0ee00
refactor entityembedding/network body ( #4857 )
4 年前
5d7345a6
Add Q function with attention
4 年前
6e6f3246
clean up
4 年前
092ea232
Some more progress - still broken
4 年前
4e37974c
moved embedding layer norm into network body
4 年前
89b6c949
use singular entity embedding ( #4873 )
4 年前
457b2630
I think it's running
4 年前
5caaef52
move entity max computation to network body
4 年前
65b866b0
Actions added but untested
4 年前
86d4c5c5
move embedding norm into RSA block
4 年前
3e481f7d
Fix issue with team_actions
4 年前
457fb612
Merge branch 'master' of https://github.com/Unity-Technologies/ml-agents
4 年前
0919a32d
Add next action and next team obs
4 年前
44088167
separate forward into q_net and baseline
4 年前
07e92563
Merge branch 'develop-centralizedcritic-counterfact' into develop-coma2
4 年前
6e1826f8
might be right
4 年前
52b011d6
_
4 年前
03c905b2
Fix equation for entropy
4 年前
1511588d
forcing this to work
4 年前
5f9ea5ea
_
4 年前
396bc43c
Merging master
4 年前
d4455936
Merge pull request #4869 from Unity-Technologies/fix-normal-entropy
Fix equation for entropy
4 年前
a4b78d53
_
4 年前
b7c7d773
Adding some tests
4 年前
6f3ea7b8
_
4 年前
92005ea6
Better placeholder for encoded_self than None
4 年前
feb38012
add lambda return and target network
4 年前
e8d54b6f
Use hypernetwork if there is a goal
4 年前
099a51ef
removing extra line
4 年前
65c2fa7f
Making EntityEmbedding encode self and entities separately
4 年前
512482c8
Check that goal is a goal
4 年前
bd4bc66b
Merge branch 'master' into fix-numti-env-delayed-spawn
4 年前
f101393c
Better bounds for log probabilities
4 年前
77eecc6b
Merge branch 'master' into develop-att-network-integration
4 年前
1cf97635
Additional conditional experiments
4 年前
7e47f94b
addressing comments
4 年前
a4c336c2
value estimator
4 年前
d7f549f9
Run pytest on GPU ( #4865 )
* make tests device-friendly
* mark all tests in test_simple_rl
4 年前
d2526ce2
Modify CrawlerDynamic
4 年前
3c1e98ca
Update ml-agents/mlagents/trainers/torch/distributions.py
4 年前
c27a95f0
Make a self encoder before EntityEmbedding
4 年前
ff70c5c4
Merge branch 'master' into goal-conditioning-new
4 年前
2bf6737f
addressing comments
4 年前
b8e81b00
Make lists modulelists
4 年前
fd007f53
Attempting to use EntityEmbedding directly as processor
4 年前
efa5a164
addressing comments and adding the changes to rpc_utils
4 年前
759fd2b5
PushJump modifications
4 年前
19a8244b
[skip ci] Removing no-longer-useful comment
We concatenate the encodes instead of adding them up
4 年前
21aaa5fe
Add goal to hyper input
4 年前
a180dbf7
Add visual version of task and simply encoders
4 年前
7f491ae7
cloud run with coma2 of held out zombie test env
4 年前
9fbc2e0e
_
4 年前
7165e9cf
Make conditiontype a setting
4 年前
9af22d30
use only value funcs
4 年前
bf16bad6
_
4 年前
511a9a7e
no baseline
4 年前
6a61eb05
-
4 年前
1acdc155
Changes to hypernet
-
-
-
experimenting
-
4 年前
82d80d72
better error message
4 年前
2fb87e4f
Merge branch 'master' into reward-dist
4 年前
04bdb40c
Reorder operations
4 年前
212ebfb9
Merge pull request #4844 from Unity-Technologies/develop-att-network-integration
Integrate attention to networkbody
4 年前
ca8a4ae2
Merge pull request #4990 from Unity-Technologies/goal-conditioning-new-vince
Tried to make improvements to the initialization of the hyperNetwork weights
4 年前
687f411b
try again on cloud
4 年前
9d70220e
Merge remote-tracking branch 'origin/master' into MLA-1734-demo-provider
4 年前
1d106816
Support multiple goals in networkbody
4 年前
7ba4ca17
Reformat line
4 年前
2552cf18
Address comments
4 年前
001990af
Merge pull request #5008 from Unity-Technologies/multi-goal-conditioning
Support multiple goals in networkbody
4 年前
f9ff3fef
shared baseline and v
4 年前
bca6c92c
Add clipping, use same network for value
4 年前
4413203d
Sensor cleanup
4 年前
3283b6a1
Remove Q-net for perf
4 年前
64fc7f43
Buffer key enums ( #4907 )
4 年前
2be83146
Use same network
4 年前
6094613d
try reduce bias more
4 年前
8baaaf4d
Changing model export to be compatible with Barracuda
4 年前
c08fefbc
reduce initialization weights
4 年前
b6f88d6d
Merge branch 'develop-base-teammanager' into develop-agentprocessor-teammanager
4 年前
eeabb974
Separate Actor/Critic, remove ActorCritics
4 年前
6bd396ee
add critic to optimizer, ppo runs
4 年前
c7054d76
Use attention tests from master
4 年前
0bde7598
Back out trainer changes
4 年前
da6a55a0
Revert "Use attention tests from master"
This reverts commit 78e052be8f36381bb6857817ff0f505716be83b9.
4 年前
7c826fb1
Working GRU
4 年前
514873bf
Use correct memories (t-1 instead of t) for training
4 年前
bac2fb68
Use attention from master
4 年前
e9025079
Properly use MemoryModule abstraction
4 年前
c74dca9f
add SharedActorCritic
4 年前
f3a2a81f
Merge branch 'develop-fix-lstms' into develop-gru
4 年前
c87bce9e
Merge branch 'master' into develop-base-teammanager
4 年前
a9116382
Bug fixes
4 年前
98d647de
MultiInputNetBody
4 年前
bd6ab0f7
Silencing a Torch warning that raises when exporting the attention module to ONNX
4 年前
9cadfa7a
Merge master -> release_13_branch-to-master
4 年前
e1b94b8b
Merge branch 'master' into develop-var-len-obs-feature
4 年前
332e9b8b
Merge pull request #4909 from Unity-Technologies/develop-var-len-obs-feature
Develop var len obs feature
4 年前
b5da488d
Merge branch 'master' into develop-base-teammanager
4 年前
dc8e8494
Merge branch 'master' into develop-critic-optimizer
4 年前
e4f51ca7
Merge remote-tracking branch 'origin/master' into MLA-1734-demo-provider
4 年前
93a59971
Merge branch 'develop-critic-optimizer' into develop-critic-op-lstm
4 年前
d4438878
Merge branch 'develop-base-teammanager' into develop-agentprocessor-teammanager
4 年前
fd3f05b9
Enable GAIL to decay
4 年前
811319c0
Use clipped gaussian distribution for entropy calculation
4 年前
7b41e5d6
Add GAIL learning rate to TB
4 年前
4d5545c8
Set ignore done=False in GAIL ( #4971 )
4 年前
c3bc8991
cleanup, don't store mask
4 年前
f409c40c
Merge branch 'master' into develop-agentprocessor-teammanager
4 年前
e46a86ad
Merge branch 'master' into develop-superpush-int
4 年前
15d512f9
Merge branch 'master' into hh/develop/dodgeball
4 年前
08db7c2f
Merge branch 'develop-agentprocessor-teammanager' into develop-coma2-trainer-mm
4 年前
3d2171c4
Add team methods to AgentAction
4 年前
9bc88c41
Running COMA (not sure if learning)
4 年前
12cef7af
Add test for GroupObs
4 年前
1fc3640e
Change AgentAction back to 0 pad and add tests
4 年前
c6904f86
Group reward function
4 年前
ffdfd8ff
Address some comments
4 年前
338af2ec
Move the Critic into the Optimizer ( #4939 )
Co-authored-by: Ervin Teng <ervin@unity3d.com>
4 年前
13fca55f
Add more comments
4 年前
60f2b72e
Rename internal function
4 年前
be45d8c0
Move padding method to AgentBufferField
4 年前
4c947151
Merge branch 'main' into hh/develop/dodgeball
4 年前
22db0335
Modified the model_serialization to have correct inputs and outputs
4 年前
61781a1a
Merge branch 'main' into develop-agentprocessor-teammanager
4 年前
5b5171f8
switching from CamelCase to snake_case
4 年前
9060da06
Merge branch 'develop-agentprocessor-teammanager' into develop-coma2-trainer
4 年前
3068ae1f
Adding a helper method for creating observation placeholder names and removed the _h and _c placeholders
4 年前
56d4c1f9
Fix to-flat and add tests
4 年前
5d517c5e
clean ups
4 年前
e2d46ca0
Merge branch 'develop-agentprocessor-teammanager' into develop-coma2-trainer
4 年前
8562471e
add inital coma optimizer tests
4 年前
c9c7e3d0
Faster NaN masking, fix masking for visual obs ( #5015 )
* Fix get mask from visual obs, large obs perf imp.
* Bug fix
* Fix typo
4 年前
c3c34267
Merge pull request #5018 from Unity-Technologies/develop-use-correct-names-for-recurrent-inputs-and-outputs
Modified the model_serialization to have correct inputs and outputs
4 年前
06c147f8
Merge remote-tracking branch 'origin/main' into goal-conditioning-new
# Conflicts:
# Project/Assets/ML-Agents/Examples/Crawler/Prefabs/CrawlerBase.prefab
# Project/Assets/ML-Agents/Examples/GridWorld/Prefabs/Area.prefab
# Project/Assets/ML-Agents/Examples/GridWorld/Scenes/GridWorld.unity
# Project/ProjectSettings/TagManager.asset
# com.unity.ml-agents/Runtime/Sensors/CameraSensor.cs
# com.unity.ml-agents/Runtime/Sensors/VectorSensor.cs
# ml-agents/mlagents/trainers/torch/networks.py
# ml-agents/mlagents/trainers/torch/utils.py
4 年前
f49dcf8e
Update network (partially)
4 年前
d36a5242
Python Dataflow for Group Manager ( #4926 )
* Make buffer type-agnostic
* Edit types of Apped method
* Change comment
* Collaborative walljump
* Make collab env harder
* Add group ID
* Add collab obs to trajectory
* Fix bug; add critic_obs to buffer
* Set group ids for some envs
* Pretty broken
* Less broken PPO
* Update SAC, fix PPO batching
* Fix SAC interrupted condition and typing
* Fix SAC interrupted again
* Remove erroneous file
* Fix multiple obs
* Update curiosity reward provider
* Update GAIL and BC
* Multi-input network
* Some minor tweaks but still broken
* Get next critic observations into value estimate
* Temporarily disable exporting
* Use Vince's ONNX export code
* Cleanup
* Add walljump collab YAML
* Lower max height
* Update prefab
* Update prefab
* Collaborative Hallway
* Set num teammates to 2
* Add config and group ids to HallwayCollab
* Fix bug with hallway collab
* E...
4 年前
fd0dd35c
Merge branch 'main' into develop-coma2-trainer
4 年前
ce1d3d88
Resolve conflicts in networkbody
4 年前
c8137dcd
Merge branch 'main' into develop-superpush-int
4 年前
fc1ee931
Fix hypernetwork
4 年前
0ddfbfee
V2 staging new model version ( #5080 )
* Make modelCheck have flavors of error messages
* ONNX exporter v3
* Using a better CheckType and a switch statement
* Removing unused message
* More tests
* Use an enum for valid versions and use GetVersion on model directly
* Maybe the model export version a static constant in Python
* Use static constructor for FailedCheck
* Use static constructor for FailedCheck
* Modifying the docstrings
* renaming LegacyDiscreteActionOutputApplier
* removing testing code
* better warning message
* Nest the CheckTypeEnum into the FailedCheck class
* Update com.unity.ml-agents/Runtime/Inference/BarracudaModelParamLoader.cs
* Adding a line explaining that legacy tensor checks are for versions 1.X only
* Modifying the changelog
* Exporting all the branches size instead of omly the sum (#5092 )
* addressing comments
* Update com.unity.ml-agents/Runtime/Inference/BarracudaModelParamLoader.cs
Co-authored...
4 年前
d1acc658
V2 staging new model version ( #5080 )
* Make modelCheck have flavors of error messages
* ONNX exporter v3
* Using a better CheckType and a switch statement
* Removing unused message
* More tests
* Use an enum for valid versions and use GetVersion on model directly
* Maybe the model export version a static constant in Python
* Use static constructor for FailedCheck
* Use static constructor for FailedCheck
* Modifying the docstrings
* renaming LegacyDiscreteActionOutputApplier
* removing testing code
* better warning message
* Nest the CheckTypeEnum into the FailedCheck class
* Update com.unity.ml-agents/Runtime/Inference/BarracudaModelParamLoader.cs
* Adding a line explaining that legacy tensor checks are for versions 1.X only
* Modifying the changelog
* Exporting all the branches size instead of omly the sum (#5092 )
* addressing comments
* Update com.unity.ml-agents/Runtime/Inference/BarracudaModelParamLoader.cs
Co-authored-by: Chris Elion <chris.elion@unity3d.co...
4 年前
81524ee8
lstm almost runs
4 年前
f16ce486
Update v2-staging from main (March 15) ( #5123 )
4 年前
921ba4f0
Update v2-staging from main (March 15) ( #5123 )
4 年前
ba2af269
[coma2] Make group extrinsic reward part of extrinsic ( #5033 )
* Make group extrinsic part of extrinsic
* Fix test and init
* Fix tests and bug
* Add baseline loss to TensorBoard
4 年前
a362b3d9
Merge remote-tracking branch 'origin/v2-staging' into csharp-obs-spec
4 年前
6ae8ea1e
[coma2] Add support for variable length obs in COMA2 ( #5038 )
* Make group extrinsic part of extrinsic
* Fix test and init
* Fix tests and bug
* Add baseline loss to TensorBoard
* Add support for variable len obs in COMA2
* Remove weird merge artifact
* Make agent action run
* Fix __getitem__ replace with slice
* Revert "Fix __getitem__ replace with slice"
This reverts commit 87a2c9d9a9342a7d2be4e9f620d1294a5c3bf22c.
* Revert "Make agent action run"
This reverts commit 59531f3746c58d62cf52f58a88e27a3e428e8946.
4 年前
ebe45056
Merge branch 'main' into release_14_branch-to-main
4 年前
0afe5f24
add slice function to agent action
4 年前
d2635e58
Action slice ( #5047 )
* add slice function to agent action
* add type/docstring to slice
* add test
4 年前
cb13a8ca
add type/docstring to slice
4 年前
f165bfb5
update comment
4 年前
970f1d40
Merge remote-tracking branch 'origin/v2-staging' into MLA-1634-ObservationSpec
4 年前
8f35bdd3
POCA trainer ( #5005 )
Co-authored-by: Ervin Teng <ervin@unity3d.com>
Co-authored-by: Ruo-Ping Dong <ruoping.dong@unity3d.com>
Co-authored-by: Chris Elion <chris.elion@unity3d.com>
Co-authored-by: Vincent-Pierre BERGES <vincentpierre@unity3d.com>
4 年前
9e77d7e1
Merge branch 'main' into develop-soccer-groupman
4 年前
e79d8a9d
[bug-fix] Move POCA critic to default device ( #5124 )
* Move critic to default device
* Make sure to clone onto default device
* Add some debug stuff
* Some more debug
* Fix issue
* Fix bool tensor too
4 年前
e6143a83
[bug-fix] Move POCA critic to default device ( #5124 ) ( #5131 )
* Move critic to default device
* Make sure to clone onto default device
* Add some debug stuff
* Some more debug
* Fix issue
* Fix bool tensor too
4 年前
ef3d6e0d
Adding Hypernetwork modules and unit tests ( #5141 )
4 年前
41dd16e8
Merge branch 'main' into release_15_mm
4 年前
90872d1d
V2 staging new model version ( #5080 )
* Make modelCheck have flavors of error messages
* ONNX exporter v3
* Using a better CheckType and a switch statement
* Removing unused message
* More tests
* Use an enum for valid versions and use GetVersion on model directly
* Maybe the model export version a static constant in Python
* Use static constructor for FailedCheck
* Use static constructor for FailedCheck
* Modifying the docstrings
* renaming LegacyDiscreteActionOutputApplier
* removing testing code
* better warning message
* Nest the CheckTypeEnum into the FailedCheck class
* Update com.unity.ml-agents/Runtime/Inference/BarracudaModelParamLoader.cs
* Adding a line explaining that legacy tensor checks are for versions 1.X only
* Modifying the changelog
* Exporting all the branches size instead of omly the sum (#5092 )
* addressing comments
* Update com.unity.ml-agents/Runtime/Inference/BarracudaModelParamLoader.cs
Co-authored...
4 年前
e83c350a
Fixing networks.py for the merge
4 年前
9609c79c
Fixing networks.py for the merge
4 年前
ea2b2f20
Merge pull request #5158 from Unity-Technologies/v2-staging-rebase-2-fix-networks
Fixing networks.py for the merge
4 年前
78851829
Merge pull request #5157 from Unity-Technologies/v2-staging-rebase-2
V2 Staging back to main
4 年前
18be47e8
Merge branch 'main' into develop-soccer-groupman-mod
4 年前
81705d6d
Goal conditioning integration ( #5142 )
* Adding Hypernetwork modules and unit tests
* Edits
* Integration of the hypernetowrk to the trainer
* Update ml-agents/mlagents/trainers/torch/networks.py
Co-authored-by: Arthur Juliani <awjuliani@gmail.com>
* Making the default hyper and added the conditioning type None
* Reducing the number of hypernetwork layers
* addressing comments
Co-authored-by: Arthur Juliani <awjuliani@gmail.com>
4 年前
d4716caa
Merge branch 'main' into goal-conditioning-sensors-3
4 年前
c37cfac1
Adding the goal conditioning sensors with the new observation specs ( #5159 )
* Fixing networks.py for the merge
* fix compile error
* Adding the goal conditioning sensors with the new observation specs
* addressing feedback
* I forgot to change the m_observationType
* Renaming Goal to GoalSignal (#5190 )
* Renaming GOAL to GOAL_SIGNAL
* VectorSensorComponent to use new API
* Adding docstrings
* verbose pytest on github action
Co-authored-by: Chris Elion <chris.elion@unity3d.com>
4 年前
1b4fd8fb
Renaming GOAL to GOAL_SIGNAL
4 年前
b9cab453
[perf] Optimizations for performance ( #5192 )
* Lazy init the buffer when sampling
* Update references rather than copy data
* Don't create unneeded numpy arrays
* Remove self[key] from loop
4 年前
2e19759c
Turning some logger.info into logger.debug and remove some logging overhead when not using debug ( #5211 )
* turning some logger.info into logger.debug and remove some logging overhead when not using debug
* Addressing comments
* Adding to changelog
4 年前
dffc37bf
Update to barracuda 1.3.3 and changes to the model inputs and outputs for LSTM ( #5236 )
* Initial commit
* making it work with hallway. Added a new model version
* addressing feedback
* Adding a comment
* formatting
* modifying the changelog
* Adding descriptions on the model version descriptions
* Update com.unity.ml-agents/Runtime/Inference/GeneratorImpl.cs
Co-authored-by: Chris Elion <chris.elion@unity3d.com>
* addressing comments
Co-authored-by: Chris Elion <chris.elion@unity3d.com>
4 年前
51adab1c
Fix the attention module embedding size
4 年前
353b1566
Fix the attention module embedding size ( #5272 )
* Fix the attention module embedding size
* editing the changelog
4 年前
1678be1c
POCA Attention will use h_size for embedding size and not 128 ( #5281 )
4 年前
5e25015f
POCA Attention will use h_size for embedding size and not 128 ( #5281 ) ( #5287 )
4 年前
4e14879d
Updating the barracuda 1.4.0 ( #5291 )
Initial commit
second commit. The no-extrinsic was trained without the log reward (reward = prob) while the new one is (reward = log_prob - log_prior)
A few results, it looks like Walker-diverse-r05-bigger.onnx is doing something
Modified pushblock
using next state and action. Did not help
Fixing bug that had 9 diversity settings instead of 8
removing results
4 年前
bf8acbb0
-
4 年前
4846cf0f
Merge branch 'main' into develop-api-documentation-update
Updating with main.
4 年前
806f04bd
Readding the validation of the minimal cnn input size ( #5345 ) ( #5346 )
4 年前
15440c24
Readding the validation of the minimal cnn input size ( #5345 )
4 年前
bb07eb45
Adding a fully connected visual encoder for super small visual input + tests ( #5351 )
* initial commit for a fully connected visual encoder
* adding a test
* addressing comments
* Fixing error with minimal size of fully connected network
* adding documentation and changelog
4 年前
ce1808f2
[bug-fix] Make resnet barracuda-compatible ( #5358 ) ( #5364 )
Co-authored-by: Ervin T <ervin@unity3d.com>
4 年前
97b7d5c6
Merge branch 'main' into develop-api-documentation-update
Syncing with main.
4 年前
20c0c728
[bug-fix] Make resnet barracuda-compatible ( #5358 )
4 年前
983982ee
Removing misleading learning rate
3 年前
Ervin Teng
GitHub
vincentpierre
Andrew Cohen
Ruo-Ping Dong
Anupam Bhatnagar
HH
Arthur Juliani
Chris Elion
brccabral
Christopher Goy
Miguel Alonso Jr