forked from facebookresearch/mtrl
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathconfigs
149 lines (112 loc) · 8.34 KB
/
configs
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
MT-50 - CARE-PTSL 200k
PYTHONPATH=. python3 -u main.py setup=metaworld env=metaworld-mt50 agent=state_sac \
agent.actor.hidden_dim=274 agent.actor.num_layers=3 experiment.num_eval_episodes=10 \
experiment.num_train_steps=200000 setup.seed_ref=1 setup.num_seeds=10 \
setup.name=PAL_shared replay_buffer.batch_size=1280 agent.encoder.type_to_select=moe \
agent.encoder.moe.task_id_to_encoder_id_cfg.mode=attention agent.encoder.moe.num_experts=4 \
agent.multitask.num_envs=50 agent.multitask.should_use_disentangled_alpha=True \
agent.multitask.should_use_task_encoder=True agent.multitask.should_use_multi_head_policy=False \
agent.multitask.task_encoder_cfg.model_cfg.pretrained_embedding_cfg.should_use=True \
agent.multitask.should_use_pal=True agent.multitask.pal_cfg.pal_dim=32 \
agent.multitask.pal_cfg.shared_projection=True agent.multitask.pal_cfg.residual_mode=none
MT-50 - CARE-PTSL 1M
PYTHONPATH=. python3 -u main.py setup=metaworld env=metaworld-mt50 agent=state_sac \
agent.actor.hidden_dim=274 agent.actor.num_layers=3 experiment.num_eval_episodes=10 \
experiment.num_train_steps=1000000 setup.seed_ref=1 setup.num_seeds=4 \
setup.name=PAL_shared replay_buffer.batch_size=1280 agent.encoder.type_to_select=moe \
agent.encoder.moe.task_id_to_encoder_id_cfg.mode=attention agent.encoder.moe.num_experts=4 \
agent.multitask.num_envs=50 agent.multitask.should_use_disentangled_alpha=True \
agent.multitask.should_use_task_encoder=True agent.multitask.should_use_multi_head_policy=False \
agent.multitask.task_encoder_cfg.model_cfg.pretrained_embedding_cfg.should_use=True \
agent.multitask.should_use_pal=True agent.multitask.pal_cfg.pal_dim=32 \
agent.multitask.pal_cfg.shared_projection=True agent.multitask.pal_cfg.residual_mode=none
MT-50 - CARE 200k
PYTHONPATH=. python3 -u main.py setup=metaworld env=metaworld-mt50 agent=state_sac \
agent.actor.hidden_dim=400 agent.actor.num_layers=3 agent.multitask.num_envs=10 \
experiment.num_eval_episodes=10 experiment.num_train_steps=200000 setup.seed_ref=1 \
setup.num_seeds=10 setup.name=CARE replay_buffer.batch_size=1280 agent.encoder.type_to_select=moe \
agent.encoder.moe.task_id_to_encoder_id_cfg.mode=attention agent.encoder.moe.num_experts=4 \
agent.multitask.num_envs=50 agent.multitask.should_use_disentangled_alpha=True \
agent.multitask.should_use_task_encoder=True agent.multitask.should_use_multi_head_policy=False \
agent.multitask.task_encoder_cfg.model_cfg.pretrained_embedding_cfg.should_use=True
MT-50 - CARE 1M
PYTHONPATH=. python3 -u main.py setup=metaworld env=metaworld-mt50 agent=state_sac \
agent.actor.hidden_dim=400 agent.actor.num_layers=3 agent.multitask.num_envs=10 \
experiment.num_eval_episodes=10 experiment.num_train_steps=1000000 setup.seed_ref=1 \
setup.num_seeds=4 setup.name=CARE replay_buffer.batch_size=1280 agent.encoder.type_to_select=moe \
agent.encoder.moe.task_id_to_encoder_id_cfg.mode=attention agent.encoder.moe.num_experts=4 \
agent.multitask.num_envs=50 agent.multitask.should_use_disentangled_alpha=True \
agent.multitask.should_use_task_encoder=True agent.multitask.should_use_multi_head_policy=False \
agent.multitask.task_encoder_cfg.model_cfg.pretrained_embedding_cfg.should_use=True
SoftMod - 200k
PYTHONPATH=. python3 -u main.py setup=metaworld env=metaworld-mt50 agent=state_sac \
experiment.num_eval_episodes=10 experiment.num_train_steps=200000 setup.seed_ref=1 \
setup.num_seeds=10 setup.name=soft_modularization replay_buffer.batch_size=1280 \
agent.multitask.num_envs=50 agent.multitask.should_use_disentangled_alpha=True \
agent.multitask.should_use_task_encoder=True agent.encoder.type_to_select=feedforward \
agent.multitask.actor_cfg.should_condition_model_on_task_info=True \
agent.multitask.actor_cfg.should_condition_encoder_on_task_info=False \
agent.multitask.actor_cfg.should_concatenate_task_info_with_encoder=False \
agent.multitask.actor_cfg.moe_cfg.should_use=True \
agent.multitask.actor_cfg.moe_cfg.mode=soft_modularization \
agent.multitask.should_use_multi_head_policy=False \
agent.encoder.feedforward.hidden_dim=50 agent.encoder.feedforward.num_layers=2 \
agent.encoder.feedforward.feature_dim=50 agent.actor.num_layers=4 \
agent.multitask.task_encoder_cfg.model_cfg.pretrained_embedding_cfg.should_use=False
SoftMod - 1M
PYTHONPATH=. python3 -u main.py setup=metaworld env=metaworld-mt50 agent=state_sac \
experiment.num_eval_episodes=10 experiment.num_train_steps=1000000 setup.seed_ref=1 \
setup.num_seeds=4 setup.name=soft_modularization replay_buffer.batch_size=1280 \
agent.multitask.num_envs=50 agent.multitask.should_use_disentangled_alpha=True \
agent.multitask.should_use_task_encoder=True agent.encoder.type_to_select=feedforward \
agent.multitask.actor_cfg.should_condition_model_on_task_info=True \
agent.multitask.actor_cfg.should_condition_encoder_on_task_info=False \
agent.multitask.actor_cfg.should_concatenate_task_info_with_encoder=False \
agent.multitask.actor_cfg.moe_cfg.should_use=True \
agent.multitask.actor_cfg.moe_cfg.mode=soft_modularization \
agent.multitask.should_use_multi_head_policy=False \
agent.encoder.feedforward.hidden_dim=50 agent.encoder.feedforward.num_layers=2 \
agent.encoder.feedforward.feature_dim=50 agent.actor.num_layers=4 \
agent.multitask.task_encoder_cfg.model_cfg.pretrained_embedding_cfg.should_use=False
MT-SAC - 200k
PYTHONPATH=. python3 -u main.py setup=metaworld env=metaworld-mt50 agent=state_sac \
agent.actor.hidden_dim=400 agent.actor.num_layers=3 agent.multitask.num_envs=50 \
experiment.num_eval_episodes=10 experiment.num_train_steps=200000 setup.seed_ref=1 \
setup.num_seeds=10 setup.name=SAC replay_buffer.batch_size=1280 \
agent.encoder.type_to_select=identity agent.multitask.num_envs=50 \
agent.multitask.should_use_disentangled_alpha=True agent.multitask.should_use_multi_head_policy=False \
agent.multitask.actor_cfg.should_condition_model_on_task_info=False \
agent.multitask.actor_cfg.should_condition_encoder_on_task_info=True \
agent.multitask.actor_cfg.should_concatenate_task_info_with_encoder=True
MT-SAC - 1M
PYTHONPATH=. python3 -u main.py setup=metaworld env=metaworld-mt50 agent=state_sac \
agent.actor.hidden_dim=400 agent.actor.num_layers=3 agent.multitask.num_envs=50 \
experiment.num_eval_episodes=10 experiment.num_train_steps=1000000 setup.seed_ref=1 \
setup.num_seeds=4 setup.name=SAC replay_buffer.batch_size=1280 \
agent.encoder.type_to_select=identity agent.multitask.num_envs=50 \
agent.multitask.should_use_disentangled_alpha=True agent.multitask.should_use_multi_head_policy=False \
agent.multitask.actor_cfg.should_condition_model_on_task_info=False \
agent.multitask.actor_cfg.should_condition_encoder_on_task_info=True \
agent.multitask.actor_cfg.should_concatenate_task_info_with_encoder=True
MT-50 - CARE-PTSL 200k (2 layers)
PYTHONPATH=. python3 -u main.py setup=metaworld env=metaworld-mt50 agent=state_sac \
agent.actor.hidden_dim=331 agent.actor.num_layers=2 experiment.num_eval_episodes=10 \
experiment.num_train_steps=200000 setup.seed_ref=1 setup.num_seeds=10 \
setup.name=PAL_shared replay_buffer.batch_size=1280 agent.encoder.type_to_select=moe \
agent.encoder.moe.task_id_to_encoder_id_cfg.mode=attention agent.encoder.moe.num_experts=4 \
agent.multitask.num_envs=50 agent.multitask.should_use_disentangled_alpha=True \
agent.multitask.should_use_task_encoder=True agent.multitask.should_use_multi_head_policy=False \
agent.multitask.task_encoder_cfg.model_cfg.pretrained_embedding_cfg.should_use=True \
agent.multitask.should_use_pal=True agent.multitask.pal_cfg.pal_dim=42 \
agent.multitask.pal_cfg.shared_projection=True agent.multitask.pal_cfg.residual_mode=none
MT-50 - CARE-PTSL 1M (2 layers)
PYTHONPATH=. python3 -u main.py setup=metaworld env=metaworld-mt50 agent=state_sac \
agent.actor.hidden_dim=331 agent.actor.num_layers=2 experiment.num_eval_episodes=10 \
experiment.num_train_steps=1000000 setup.seed_ref=1 setup.num_seeds=4 \
setup.name=PAL_shared replay_buffer.batch_size=1280 agent.encoder.type_to_select=moe \
agent.encoder.moe.task_id_to_encoder_id_cfg.mode=attention agent.encoder.moe.num_experts=4 \
agent.multitask.num_envs=50 agent.multitask.should_use_disentangled_alpha=True \
agent.multitask.should_use_task_encoder=True agent.multitask.should_use_multi_head_policy=False \
agent.multitask.task_encoder_cfg.model_cfg.pretrained_embedding_cfg.should_use=True \
agent.multitask.should_use_pal=True agent.multitask.pal_cfg.pal_dim=42 \
agent.multitask.pal_cfg.shared_projection=True agent.multitask.pal_cfg.residual_mode=none