issues
search
instadeepai
/
Mava
🦁 A research-friendly codebase for fast experimentation of multi-agent reinforcement learning in JAX
Apache License 2.0
709
stars
83
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
feat: add ability to arbitrarily chain torso configs together
#1099
EdanToledo
opened
2 weeks ago
0
[FEATURE] Use numpyro
#1098
sash-a
opened
1 month ago
0
feat: add grad clipping to sac
#1097
sash-a
closed
1 month ago
0
fix: sac key splitting
#1096
sash-a
closed
1 month ago
0
chore: change all jax tree_map to tree.map
#1095
WiemKhlifi
closed
2 months ago
0
Chore: sebulba arch update
#1094
Louay-Ben-nessir
closed
2 months ago
0
feat: upgrade supported python version
#1093
sash-a
closed
2 months ago
0
[MAINTAIN] Use uv for managing dependencies.
#1092
RuanJohn
opened
2 months ago
0
[FEATURE] GNN support for the MARL frameworks
#1091
Jaroan
opened
2 months ago
1
Chore: anakin and sebulba folders
#1090
Louay-Ben-nessir
closed
2 months ago
1
fix: unpack array shape when reshaping minibatches in mappo training
#1089
RuanJohn
closed
2 months ago
0
Feat: sebulba ff_ippo
#1088
Louay-Ben-nessir
opened
2 months ago
0
fix: remove weak types to prevent double compilation
#1087
EdanToledo
closed
2 months ago
2
[BUG] Double compilation of learn_fn
#1086
EdanToledo
closed
2 months ago
0
feat: API for all Anakin Mava Envs
#1085
sash-a
closed
2 months ago
0
feat: refactor evaluator
#1084
sash-a
closed
2 months ago
0
chore: changed the assertions to make sure the num_updates is a multiple of num_evaluations
#1083
Louay-Ben-nessir
opened
3 months ago
3
[BUG] LLVM ERROR: mma16816 data type not supported
#1082
UsaidPro
closed
2 months ago
3
feat: ruff formatter and linter
#1081
sash-a
closed
2 months ago
0
feat: gym wrapper for sebulba
#1080
Louay-Ben-nessir
closed
2 months ago
0
[FEATURE] Support for the Sebulba architecture
#1079
Louay-Ben-nessir
opened
3 months ago
0
[BUG] steps_per_second under-reporting in SAC and IQL
#1078
JemmaLDaniel
opened
4 months ago
0
Chore: Connector Update
#1077
SimonDuToit
closed
2 months ago
2
fix: fix off by one error in rnn dones used in the transition
#1076
RuanJohn
closed
4 months ago
0
fix/rec-iql and sac timestep logging
#1075
JemmaLDaniel
closed
4 months ago
0
[BUG] Quickstart Colab imports failing
#1074
Brunozml
closed
5 months ago
2
fix: pin mujoco version for jaxmarl to work
#1073
RuanJohn
closed
5 months ago
0
fix: pin the scipy version to 1.12.0 to prevent breaking changes introduced in 1.13.0
#1072
liamclarkza
closed
6 months ago
0
Feature: Control number of vmapped envs in evaluator using `arch.num_envs`
#1071
OmaymaMahjoub
closed
3 months ago
1
feat: integration tests
#1070
sash-a
closed
2 months ago
0
feat: separate config option for logging winrate
#1069
sash-a
closed
6 months ago
0
feat: allow scenarios to set env kwargs
#1068
sash-a
closed
6 months ago
0
feat: cnn support for recurrent systems
#1067
SimonDuToit
closed
3 months ago
0
fix: final return value for SAC systems
#1066
sash-a
closed
6 months ago
0
Make sure that config params passed to fbx.make_item_buffer are ints
#1065
liamclarkza
closed
6 months ago
0
fix: quickstart notebook
#1064
WiemKhlifi
closed
5 months ago
2
Chore: Rename jax file under utils folder
#1063
WiemKhlifi
closed
6 months ago
0
Jumanji Wrapper: add a cast from Jumanji Observation to Mava Observation for consistent types in the observation_spec function
#1062
lbeyers
closed
6 months ago
0
Recurrent IQL
#1061
lbeyers
closed
6 months ago
3
[FEATURE] Set time limit per scenario
#1060
SimonDuToit
closed
6 months ago
1
feat: uncouple hiddenstate form previous layer in recurrent policies
#1059
RuanJohn
closed
6 months ago
0
[FEATURE] ScannedRNN hidden state initialisation improvement
#1058
lbeyers
closed
6 months ago
0
feat: cleaner wrapper
#1057
SimonDuToit
closed
6 months ago
0
chore: merge updated mava
#1056
SimonDuToit
closed
7 months ago
0
Generalise the evaluator factory to accept a custom apply function instead of enforcing network.apply
#1055
lbeyers
closed
6 months ago
2
fix: steps per second count of the evaluator
#1054
OmaymaMahjoub
closed
7 months ago
0
fix: fix CNN torso to maintain batch and agent dimension
#1053
RuanJohn
closed
7 months ago
0
feat: mkdocs documentation
#1052
liamclarkza
opened
7 months ago
1
New mava network: epsilon greedy action head for discrete action spaces
#1051
lbeyers
closed
6 months ago
2
docs: update readme with continuous system details
#1050
RuanJohn
opened
7 months ago
0
Next