issues
search
facebookresearch
/
rlmeta
RLMeta is a light-weight flexible framework for Distributed Reinforcement Learning Research.
MIT License
284
stars
28
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Modify GAE op interface
#116
xiaomengy
closed
1 year ago
0
Add C++ implementation of GAE op
#115
xiaomengy
closed
1 year ago
0
Add C++ implementation of discounted_return op.
#114
xiaomengy
closed
1 year ago
0
Enable eval mode for Atari Ape-X DQN example
#113
xiaomengy
closed
1 year ago
0
Add stats_dict_test
#112
xiaomengy
closed
1 year ago
0
Improve Ape-X DQN Implementation
#111
xiaomengy
closed
1 year ago
0
Add stacked parameter for ReplayBuffer.extend to support stacked tensor
#110
xiaomengy
closed
1 year ago
0
Minor change in CONTRIBUTING.md
#109
xiaomengy
closed
1 year ago
0
Refactor Atari Models and Atari Game settings
#108
xiaomengy
closed
1 year ago
0
Improve Atari Examples
#107
xiaomengy
closed
1 year ago
0
Fix LazyFrames observation
#106
xiaomengy
closed
1 year ago
0
Update README.md to be aligned with new AtariEnv
#105
xiaomengy
closed
1 year ago
0
Deprecate old atari_wrappers
#104
xiaomengy
closed
1 year ago
0
Fix bugs when reward is int type
#103
xiaomengy
closed
1 year ago
0
Marl dev
#102
cuijiaxun
closed
1 year ago
0
Switch to the new OpenAI Gym APIs
#101
xiaomengy
closed
1 year ago
0
Update Ape-X DQN implementation with tricks in MEME
#100
xiaomengy
closed
1 year ago
0
Use torch.float64 for running_stats in Rescalers to reduce numeric error
#99
xiaomengy
closed
1 year ago
0
improve ModelPool implementation to support different model versions
#98
xiaomengy
closed
1 year ago
0
Tiny improvement for ModelPool
#97
xiaomengy
closed
1 year ago
0
Fix bugs in model_pool
#96
xiaomengy
closed
1 year ago
0
Add max_abs_reward when rescaling rewards for PPO
#95
xiaomengy
closed
1 year ago
0
Add ModelPool to support different version of models
#94
xiaomengy
closed
1 year ago
0
Add groupby op to support multi-version model inference
#93
xiaomengy
closed
1 year ago
0
Add at/get/front/back APIs for CircularBuffer
#92
xiaomengy
closed
1 year ago
0
Add loss_utils to support easy config for loss functions
#91
xiaomengy
closed
1 year ago
0
Fix reward scaling implementation in PPO
#90
xiaomengy
closed
1 year ago
0
Add replacement argument to ReplayBuffer.sample
#89
xiaomengy
closed
1 year ago
0
Pip installation fails in virtual env and SIGILL on DGX machines
#88
d3sm0
opened
1 year ago
0
TensorCircularBuffer with capacity larger of 1mln fails
#87
d3sm0
opened
1 year ago
1
Fix bug in TensorCircularBuffer::Clear()
#86
xiaomengy
closed
1 year ago
0
Replay buffer crashes after being cleared
#85
d3sm0
closed
1 year ago
4
Slightly update DQN implementation
#84
xiaomengy
closed
1 year ago
0
How to sample partial trajectories?
#83
d3sm0
closed
1 year ago
1
Refactor Ape-X DQN Agent
#82
xiaomengy
closed
1 year ago
0
Improve PPO implementation to avoid potential timeout for RPC
#81
xiaomengy
closed
1 year ago
0
Refactor PPOAgent and PPORNDAgent
#80
xiaomengy
closed
1 year ago
0
Add optimizer_utils to config optimizers in hydra
#79
xiaomengy
closed
1 year ago
0
Switch to new OpenAI Gym step API
#78
xiaomengy
closed
1 year ago
2
Add StdRescaler
#77
xiaomengy
closed
1 year ago
0
Minor fix for RMSRescaler to avoid NaN
#76
xiaomengy
closed
1 year ago
0
Fix function definitions of DQN
#75
xiaomengy
closed
1 year ago
0
Refactor ReplayBuffer to improve flexibility and performance
#74
xiaomengy
closed
2 years ago
0
[WIP] Refactor ReplayBuffer to improve flexibility and performance.
#73
xiaomengy
closed
2 years ago
0
Enable training data collection during evaluation phase
#72
xiaomengy
closed
2 years ago
0
Improve ApeX-DQN implementation
#71
xiaomengy
closed
2 years ago
0
[WIP] Refactor ReplayBuffer Design and Implementation
#70
xiaomengy
closed
2 years ago
0
Sync moolib submodule to lastest version
#69
xiaomengy
closed
2 years ago
0
Switch to latest version of moolib
#68
xiaomengy
closed
2 years ago
0
Improve SegmentTree space usage
#67
xiaomengy
closed
2 years ago
0
Next