search

pengyanghua / DL2

a deep learning-driven scheduler for elastic training in deep learning clusters
27 stars 4 forks source link

train.py is an implementation of A3C or A2C ? #6

Open wanziyu opened 2 years ago

wanziyu commented 2 years ago

In train.py, I see a central agent，SL agent and RL agents. They are running in different CPU cores with multiprocessing package. And RL agents get the weights of policy and value network from central agent with a Queue. I see train_a3c.py is very similar to train.py. I wonder if these two files are both implementations of A3C algorithm?

pengyanghua commented 2 years ago

@wanziyu I think both are A3C algorithm. You can run "diff train.py train_a3c.py" to see the differences.