How to evaluate and test the training result?

shariqiqbal2810 / MAAC

Code for "Actor-Attention-Critic for Multi-Agent Reinforcement Learning" ICML 2019

MIT License

676 stars 173 forks source link

How to evaluate and test the training result? #12

Closed wanghuimu closed 5 years ago

wanghuimu commented 5 years ago

Hi, Your codes are written very well~ Thanks for your great job! But in all folders, I only found codes for training the model and I didn’t find the codes for testing or evaluating the trained model in environments , would you mind uploading the evaluation codes file for testing or evaluating? Thank you~

wang88256187 commented 5 years ago

Hi, Your codes are written very well~ Thanks for your great job! But in all folders, I only found codes for training the model and I didn’t find the codes for testing or evaluating the trained model in environments , would you mind uploading the evaluation codes file for testing or evaluating? Thank you~

hi, Have you got the final result ? why did I always meet the issue that the machine memorry always overflow? Can you give me some reason?

niniuba123456 commented 5 years ago

Hi, Your codes are written very well~ Thanks for your great job! But in all folders, I only found codes for training the model and I didn’t find the codes for testing or evaluating the trained model in environments , would you mind uploading the evaluation codes file for testing or evaluating? Thank you~

Did you solve this problem?

shariqiqbal2810 commented 5 years ago

Hi,

I do not have the bandwidth to release evaluation code at the moment. The process for evaluating a trained model is as such:

Load the saved model file using AttentionSAC.init_from_save(<filename>)
Rollout in the environment using the appropriate number of maximum steps (detailed in README)
Keep track of total rewards per episode per agent
Average episodic rewards (across Rovers for Rover-Tower and across Collectors for CTC) and report these values.