search

TianhongDai / hindsight-experience-replay

This is the pytorch implementation of Hindsight Experience Replay (HER) - Experiment on all fetch robotic environments.
MIT License
402 stars 75 forks source link

FetchPickAndPlace-v1 #18

Open quyouyuan opened 3 years ago

quyouyuan commented 3 years ago

Hello! I'm sorry to bother you again. I created a sub optimal policy when running ' mpirun -oversubscribe -np 16 python -u train.py --env-name='FetchPickAndPlace-v1' 2>&1 | tee pick.log'. When the success rate reaches 0.9, it will not increase. Can you analyze the reason? Thank you very much! I'm sorry to bother you again!

2021-09-30 10-21-47 的屏幕截图

TianhongDai commented 3 years ago

@quyouyuan No worries! In actually, that is the problem of HER itself. The performance of HER on the PickAndPlace task is only around 90%, please refer to Fig.3 in https://arxiv.org/pdf/1802.09464.pdf (this is the OpenAI's report of HER). You may expect to propose some new methods to improve the performance of HER. Furthermore, I also find that it can improve the performance slightly (around 94%) by sampling diverse goals: https://arxiv.org/pdf/2108.07887.pdf.

123gzy commented 3 weeks ago

请问环境呢 我怎么找不到环境