Mava icon indicating copy to clipboard operation
Mava copied to clipboard

🦁 A research-friendly codebase for fast experimentation of multi-agent reinforcement learning in JAX

Results 160 Mava issues
Sort by recently updated
recently updated
newest added

### Feature Add a feature of death masking by masking the loss so that we have 3 options when we use SMAC: 1. Without death masking 2. Death masking by...

enhancement
maintenance

### What do you want to investigate? Make sure we are following best practices when running docker containers in our CI.

maintenance

### Please describe what needs to be maintained? Unify the use of certain types across mava and remove our dependency on typing from other libraries such as acme as much...

maintenance

### Please describe the purpose of the feature. Is it related to a problem? Implement a method to automatically track code duplication in Mava. [Pylint](https://julien.duponchelle.info/python/detect-python-code-duplicate) has this functionality. Is it...

enhancement

### Please describe the purpose of the feature. Is it related to a problem? Implement a component that can compute and log the absolute metric for a given experiment run....

enhancement

### Feature A logger that can write data from the evaluator process to a structured json file. This will enable for the usage of the EMARL repo tools library for...

enhancement

### Describe the bug Jax example runs but gives slow compilation warnings: `[trainer/0] [Compiling module jit_sgd_step.8] Very slow compile? If you want to file a bug, run with envvar XLA_FLAGS=--xla_dump_to=/tmp/foo...

bug

### Feature Create a feature which allows a system to save its complete code to file. ### Proposal This component would provide a system with a `.compile()` functionality. When one...

enhancement

### Please describe the purpose of the feature. Is it related to a problem? Checkpointer does not currently checkpoint the JAX RNG states for each node ### Describe the solution...

enhancement

### Please describe the purpose of the feature. Is it related to a problem? Checkpointer currently checkpoints latest parameters at fixed intervals instead of checkpointing the best params/ network weights...

enhancement