From 4e1ef899249a405340502d24d412d532979dddaf Mon Sep 17 00:00:00 2001 From: VedantDave Date: Fri, 31 Mar 2023 18:00:07 +0200 Subject: [PATCH] Adding action network --- DPI/train.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/DPI/train.py b/DPI/train.py index 29d0cab..87b928d 100644 --- a/DPI/train.py +++ b/DPI/train.py @@ -48,6 +48,7 @@ def parse_args(): parser.add_argument('--state_size', default=256, type=int) parser.add_argument('--hidden_size', default=128, type=int) parser.add_argument('--history_size', default=128, type=int) + parser.add_argument('--num-units', type=int, default=200, help='num hidden units for reward/value/discount models') parser.add_argument('--load_encoder', default=None, type=str) parser.add_argument('--imagination_horizon', default=15, type=str) # eval @@ -197,7 +198,7 @@ class DPI: if args.save_video: self.env.video.init(enabled=True) self.env_clean.video.init(enabled=True) - + for i in range(self.args.episode_length): action = self.env.action_space.sample() @@ -258,7 +259,7 @@ class DPI: self.actions, self.history, i) - + print(past_encoder_loss, past_latent_loss) previous_information_loss = past_latent_loss previous_encoder_loss = past_encoder_loss