DI-engine
DI-engine copied to clipboard
opendilab
OpenDILab Decision AI Engine
## Description 1. add LeagueCoordinator, LeagueLearnerCommunicator, StepLeagueActor, BattleStepCollector, battle_inferencer, battle_rolloutor, with their corresponding tests. 2. add BattleTransitionList to gather transitions and cut trajectories for league environment 3. add dataclass of...
## Description DI-engine integrates torch.distributed.rpc module. 1. CPU-P2P-RDMA: In IB network environment, support RDMA CPU-P2P transmission 2. GPU-P2P-RDMA: supports GPU p2p communication ## Related Issue ## TODO 1. Dynamic communication...
目前`interaction`仍在使用`flask 1.1.2`，这是一个比较古老的版本，且为了确保依赖上不出问题，需要配置一系列其他的依赖（例如markupsafe、jinja2等，为此也不得不锁死为低版本）。 而实际上，对于这样的交互场景，使用web服务层级上的框架显然有些浪费（werkzueg这个层级的会好一些，当然了终极解决方案就是直接上socket）。 之前考虑到系统稳定性，初步尝试过迁移到`flask>=2`，出现了严重的问题（由于一个关键接口被删除所致，详见：https://github.com/pallets/werkzeug/issues/1752 ）。 此外，由于现在IGM的开发，同样对Jinja2版本有要求，因此可能需要针对interaction进行一次重构，以解决这个历史遗留问题。
## Description Use rainbow algorithm on mountan car environment. I intend for this branch to hold mountain car + Rainbow work. ## Related Issue ## TODO ## Check List -...
## Description ## Related Issue ## TODO ## Check List - [ ] merge the latest version source branch/repo, and resolve all the conflicts - [ ] pass style check...
## Description ## Related Issue ## TODO ## Check List - [ ] merge the latest version source branch/repo, and resolve all the conflicts - [ ] pass style check...
## Description Add adaptation to gymnasium environment and give examples based on DQN algorithm ## Related Issue ## TODO ## Check List - [ ] merge the latest version source...
## Introduction Dear community, :rocket: DI-engine aims to be a generalized decision intelligence engine, including various algorithm toolkits and application demonstrations. Due to the complexity of abstraction and maintainment, DI-engine...
This issue will keep tracking of DI-engine's updates of next few versions: ### Future - [ ] #771 - [ ] #636 - [ ] #603 - [ ] #562...