slm_lab/spec/experimental/sac/sac_cartpole.json
{
"sac_cartpole": {
"agent": [{
"name": "SoftActorCritic",
"algorithm": {
"name": "SoftActorCritic",
"action_pdtype": "GumbelSoftmax",
"action_policy": "default",
"gamma": 0.99,
"training_frequency": 1
},
"memory": {
"name": "Replay",
"batch_size": 256,
"max_size": 100000,
"use_cer": true
},
"net": {
"type": "MLPNet",
"hid_layers": [64],
"hid_layers_activation": "relu",
"init_fn": "orthogonal_",
"clip_grad_val": 0.5,
"loss_spec": {
"name": "MSELoss"
},
"optim_spec": {
"name": "Adam",
"lr": 0.005
},
"lr_scheduler_spec": null,
"update_type": "polyak",
"update_frequency": 1,
"polyak_coef": 0.005,
"gpu": false
}
}],
"env": [{
"name": "CartPole-v0",
"max_t": null,
"max_frame": 200000,
"num_envs": 8,
"normalize_state": false
}],
"body": {
"product": "outer",
"num": 1
},
"meta": {
"distributed": false,
"log_frequency": 500,
"eval_frequency": 500,
"rigorous_eval": 0,
"max_session": 1,
"max_trial": 1,
}
}
}