liruiw commited on
Commit
9d6cd99
1 Parent(s): ca19244

Upload 3 files

Browse files
Files changed (3) hide show
  1. config.yaml +21 -21
  2. model.pth +2 -2
  3. trunk.pth +2 -2
config.yaml CHANGED
@@ -1,5 +1,5 @@
1
  seed: 0
2
- output_dir: output/08_05_2024_21_50_12_1563562
3
  domains: austin_sailor_dataset_converted_externally_to_rlds, stanford_hydra_dataset_converted_externally_to_rlds,
4
  austin_buds_dataset_converted_externally_to_rlds, austin_sirius_dataset_converted_externally_to_rlds,
5
  berkeley_mvp_converted_externally_to_rlds, berkeley_rpt_converted_externally_to_rlds,
@@ -15,15 +15,15 @@ domains: austin_sailor_dataset_converted_externally_to_rlds, stanford_hydra_data
15
  columbia_cairlab_pusht_real, conq_hose_manipulation, dlr_edan_shared_control_converted_externally_to_rlds,
16
  berkeley_gnm_sac_son, berkeley_autolab_ur5, aloha_mobile, agent_aware_affordances,
17
  human_image_hammer, human_image_spatula, epic_kitchen, drake_tooluse_image, pybullet_grasping_image,
18
- mujoco_metaworld, mujoco_robomimic, isaac_arnold_image
19
- log_dir: output/08_05_2024_21_50_12_1563562
20
  debug_distributed: false
21
  wb_tag: default
22
- wb_cont_run: 13h28mb6
23
  log_interval: 10
24
- script_name: run_resnet_30dataset_traj100000_embed_512_batch2048_800k_simhuman
25
  save_wb_checkpoint: true
26
- slurm_job_id: '27327006'
27
  effective_total_epochs: 100
28
  effective_batch_size: 256
29
  epoch_size: 10
@@ -32,17 +32,17 @@ total_num_sample: 0
32
  rank: 0
33
  gpu: 0
34
  task_per_gpu: 1
35
- world_size: 32
36
  debug_submitit: false
37
  ngpus: 8
38
- nodes: 4
39
  timeout: 4320
40
  job_dir: logs/
41
  partition: learnlab
42
  use_volta32: true
43
  comment: ''
44
  resume: logs/
45
- dist_url: file:///checkpoint/xinleic/experiments/7b45bdd38afd4f85b05d324035d032b0_init
46
  dist_on_itp: false
47
  local_rank: 1
48
  distributed: true
@@ -50,7 +50,7 @@ dist_backend: nccl
50
  dset_w_temperature: 2.0
51
  dataset_shuffle: true
52
  dataset_groups: ''
53
- nodelist: learnlab,learnfair,scavenge
54
  fsdp: false
55
  count_flops: false
56
  accumuate_gradient_steps: 1
@@ -83,9 +83,9 @@ dataset:
83
  continue_generate: false
84
  network:
85
  _target_: hpt_pretrain.models.policy.Policy
86
- embed_dim: 512
87
- num_blocks: 16
88
- num_heads: 8
89
  use_modality_embedding: true
90
  use_domain_embedding: false
91
  token_postprocessing: mean
@@ -98,7 +98,7 @@ stem:
98
  modalities:
99
  - image
100
  - state
101
- modality_embed_dim: 512
102
  normalize_state: ${dataset.normalize_state}
103
  state_embedding_dim: 1
104
  image_encoder: ${dataset.image_encoder}
@@ -115,27 +115,27 @@ stem:
115
  image:
116
  _target_: hpt_pretrain.models.policy_stem.MLP
117
  input_dim: 512
118
- output_dim: 512
119
  widths:
120
  - 128
121
  num_of_copy: 1
122
  state:
123
  _target_: hpt_pretrain.models.policy_stem.MLP
124
- input_dim: 4
125
- output_dim: 512
126
  widths:
127
  - 128
128
  head:
129
  _target_: hpt_pretrain.models.policy_head.MLP
130
- input_dim: 512
131
  tanh_end: true
132
- output_dim: 56
133
  dropout: true
134
  widths:
135
  - 256
136
  - 128
137
  dataloader:
138
- batch_size: 64
139
  num_workers: 1
140
  pin_memory: false
141
  persistent_workers: false
@@ -172,7 +172,7 @@ train:
172
  epoch_iters: 1000
173
  validation_iters: 100
174
  use_accumulation: false
175
- pretrained_dir: '08_05_2024_21_50_12_1563562'
176
  max_validation_size: 10
177
  load_pretrain_trunk_only: false
178
  freeze_trunk: false
 
1
  seed: 0
2
+ output_dir: output/14_05_2024_13_09_38_3482822
3
  domains: austin_sailor_dataset_converted_externally_to_rlds, stanford_hydra_dataset_converted_externally_to_rlds,
4
  austin_buds_dataset_converted_externally_to_rlds, austin_sirius_dataset_converted_externally_to_rlds,
5
  berkeley_mvp_converted_externally_to_rlds, berkeley_rpt_converted_externally_to_rlds,
 
15
  columbia_cairlab_pusht_real, conq_hose_manipulation, dlr_edan_shared_control_converted_externally_to_rlds,
16
  berkeley_gnm_sac_son, berkeley_autolab_ur5, aloha_mobile, agent_aware_affordances,
17
  human_image_hammer, human_image_spatula, epic_kitchen, drake_tooluse_image, pybullet_grasping_image,
18
+ mujoco_metaworld, mujoco_robomimic, isaac_arnold_image, pybullet_trifinger, mujoco_adroit
19
+ log_dir: output/14_05_2024_13_09_38_3482822
20
  debug_distributed: false
21
  wb_tag: default
22
+ wb_cont_run: 2n1ryxkw
23
  log_interval: 10
24
+ script_name: run_resnet_52dataset_traj100000_embed768_batch4096_800k_32block
25
  save_wb_checkpoint: true
26
+ slurm_job_id: '27595011'
27
  effective_total_epochs: 100
28
  effective_batch_size: 256
29
  epoch_size: 10
 
32
  rank: 0
33
  gpu: 0
34
  task_per_gpu: 1
35
+ world_size: 128
36
  debug_submitit: false
37
  ngpus: 8
38
+ nodes: 16
39
  timeout: 4320
40
  job_dir: logs/
41
  partition: learnlab
42
  use_volta32: true
43
  comment: ''
44
  resume: logs/
45
+ dist_url: file:///checkpoint/xinleic/experiments/3e51596f51c04660aea04b183dba259e_init
46
  dist_on_itp: false
47
  local_rank: 1
48
  distributed: true
 
50
  dset_w_temperature: 2.0
51
  dataset_shuffle: true
52
  dataset_groups: ''
53
+ nodelist: learnlab,learnfair
54
  fsdp: false
55
  count_flops: false
56
  accumuate_gradient_steps: 1
 
83
  continue_generate: false
84
  network:
85
  _target_: hpt_pretrain.models.policy.Policy
86
+ embed_dim: 768
87
+ num_blocks: 32
88
+ num_heads: 16
89
  use_modality_embedding: true
90
  use_domain_embedding: false
91
  token_postprocessing: mean
 
98
  modalities:
99
  - image
100
  - state
101
+ modality_embed_dim: 768
102
  normalize_state: ${dataset.normalize_state}
103
  state_embedding_dim: 1
104
  image_encoder: ${dataset.image_encoder}
 
115
  image:
116
  _target_: hpt_pretrain.models.policy_stem.MLP
117
  input_dim: 512
118
+ output_dim: 768
119
  widths:
120
  - 128
121
  num_of_copy: 1
122
  state:
123
  _target_: hpt_pretrain.models.policy_stem.MLP
124
+ input_dim: 45
125
+ output_dim: 768
126
  widths:
127
  - 128
128
  head:
129
  _target_: hpt_pretrain.models.policy_head.MLP
130
+ input_dim: 768
131
  tanh_end: true
132
+ output_dim: 192
133
  dropout: true
134
  widths:
135
  - 256
136
  - 128
137
  dataloader:
138
+ batch_size: 32
139
  num_workers: 1
140
  pin_memory: false
141
  persistent_workers: false
 
172
  epoch_iters: 1000
173
  validation_iters: 100
174
  use_accumulation: false
175
+ pretrained_dir: '14_05_2024_13_09_38_3482822'
176
  max_validation_size: 10
177
  load_pretrain_trunk_only: false
178
  freeze_trunk: false
model.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a0fab4ed6b34aec993b03752e461c5ef6f32d2107de50fcda6dc718c11b5f0d1
3
- size 699884363
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac92945be3aad4149da11ca8a337ed32c36f1b18f1cc65ea8816e9f459244c92
3
+ size 1672892039
trunk.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e1e96450b017ffca43e2acd7aba7d9744087b14065ca4e1752f5cd19f29dd30e
3
- size 201883562
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:86c5583956e3f8b63e150e19a6c2e9a8faacec58d8e6fdbcd4229a1f3db31d1a
3
+ size 907575146