MarcoMeter/episodic-transformer-memory-ppo
Clean baseline implementation of PPO using an episodic TransformerXL memory
PythonMIT
Issues
- 2
Queries should be used instead of query in lines 46-47 of the transformer.cpp file.
#24 opened by LeZheng-x - 2
- 1
Inquiry on Research Publication
#20 opened by CiouQQ - 5
Question context window
#21 opened by Reytuag - 3
- 2
- 1
- 1
Question about attention when mask is all zero
#17 opened by zerlinwang - 5
Questions about the implementation
#16 opened by hai-h-nguyen - 22
- 1
Memory indices in enjoy.py
#13 opened by Reytuag - 2
Baseline results on GRU and LSTM on MemoryGym
#11 opened by subho406 - 1
Does this repo implement a WIP paper?
#10 opened by hai-h-nguyen - 4
ReLU in residual connections?
#7 opened by ibagur