Hugging Face's logo

junhalee
/
idm-source_depth_flow-bs512-it34K

Safetensors
model_hub_mixin
pytorch_model_hub_mixin
Model card Files
xet
 Community
junhalee commited on
Commit
3bce8d9
·
verified ·
1 Parent(s): 3a59eac

Upload folder using huggingface_hub

Browse files
Files changed (4) hide show
  1. README.md +10 -0
  2. config.json +105 -0
  3. model.safetensors +3 -0
  4. train_meta.json +96 -0
README.md ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - model_hub_mixin
4
+ - pytorch_model_hub_mixin
5
+ ---
6
+
7
+ This model has been pushed to the Hub using the [PytorchModelHubMixin](https://huggingface.co/docs/huggingface_hub/package_reference/mixins#huggingface_hub.PyTorchModelHubMixin) integration:
8
+ - Code: [More Information Needed]
9
+ - Paper: [More Information Needed]
10
+ - Docs: [More Information Needed]
config.json ADDED
@@ -0,0 +1,105 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cfg": {
3
+ "action_dim": 7,
4
+ "backbone": "facebook/dinov3-vits16-pretrain-lvd1689m",
5
+ "decoder_heads": 8,
6
+ "decoder_layers": 4,
7
+ "dropout": 0.1,
8
+ "encoder_input_size": 224,
9
+ "freeze_backbone": true,
10
+ "gripper_idx": 6,
11
+ "gripper_weight": 1.0,
12
+ "head_hidden": 1024,
13
+ "head_layers": 3,
14
+ "head_type": "xattn",
15
+ "huber_delta": 0.1,
16
+ "image_size": 256,
17
+ "input_mode": "source_depth_flow",
18
+ "norm_mode": "zscore",
19
+ "norm_stats": {
20
+ "actions": {
21
+ "mean": [
22
+ 0.047240227460861206,
23
+ 0.0508635975420475,
24
+ -0.09378840774297714,
25
+ 0.0036939603742212057,
26
+ 0.003270887304097414,
27
+ -0.006564674433320761,
28
+ -0.059840139001607895
29
+ ],
30
+ "q01": [
31
+ -0.46339285373687744,
32
+ -0.7392857074737549,
33
+ -0.7794643044471741,
34
+ -0.07392857223749161,
35
+ -0.16607142984867096,
36
+ -0.2978571355342865,
37
+ -1.0
38
+ ],
39
+ "q99": [
40
+ 0.7098749876022339,
41
+ 0.7821428775787354,
42
+ 0.8303571343421936,
43
+ 0.17892856895923615,
44
+ 0.1639285683631897,
45
+ 0.375,
46
+ 1.0
47
+ ],
48
+ "std": [
49
+ 0.3100294768810272,
50
+ 0.3630746901035309,
51
+ 0.4150514304637909,
52
+ 0.04578809812664986,
53
+ 0.0601838082075119,
54
+ 0.08567292243242264,
55
+ 0.9982079863548279
56
+ ]
57
+ },
58
+ "state": {
59
+ "mean": [
60
+ -0.07068316638469696,
61
+ 0.018536901101469994,
62
+ 0.8913209438323975,
63
+ 2.9518160820007324,
64
+ -0.17911545932292938,
65
+ -0.12059465050697327,
66
+ 0.02673562802374363,
67
+ -0.027061836794018745
68
+ ],
69
+ "q01": [
70
+ -0.2335965633392334,
71
+ -0.09221210330724716,
72
+ 0.9166392087936401,
73
+ 1.5836201906204224,
74
+ -2.489743947982788,
75
+ -1.0587904453277588,
76
+ 0.0018643600633367896,
77
+ -0.04007481783628464
78
+ ],
79
+ "q99": [
80
+ 0.09121209383010864,
81
+ 0.21421630680561066,
82
+ 1.1849141120910645,
83
+ 3.1840009689331055,
84
+ 3.0570921897888184,
85
+ 0.8673243522644043,
86
+ 0.04050664231181145,
87
+ -0.0014549389015883207
88
+ ],
89
+ "std": [
90
+ 0.11321546137332916,
91
+ 0.14622576534748077,
92
+ 0.2981572151184082,
93
+ 0.4244987368583679,
94
+ 1.0269798040390015,
95
+ 0.3394932448863983,
96
+ 0.014265833422541618,
97
+ 0.01417431328445673
98
+ ]
99
+ }
100
+ },
101
+ "num_actions": 8,
102
+ "proprio_dim": 8,
103
+ "proprio_hidden": 256
104
+ }
105
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d038dfef4d2c7d09e6359e7aa3fbc8b4e78214c7e171e96fceb4d8dbcd40a605
3
+ size 124921332
train_meta.json ADDED
@@ -0,0 +1,96 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "args": {
3
+ "config": "configs/idm_exp2_depth.yaml",
4
+ "data_dir": "/NHNHOME/WORKSPACE/0426030021_A/junhalee/datasets/libero_lerobot/",
5
+ "subsets": [
6
+ "libero_10_no_noops_1.0.0_lerobot",
7
+ "libero_spatial_no_noops_1.0.0_lerobot",
8
+ "libero_goal_no_noops_1.0.0_lerobot",
9
+ "libero_object_no_noops_1.0.0_lerobot",
10
+ "libero_90_no_noops_lerobot"
11
+ ],
12
+ "flow_cache_dir": "/NHNHOME/WORKSPACE/0426030021_A/junhalee/datasets/libero_flow_cache",
13
+ "flow_store": "raft_h8_r256.zarr",
14
+ "depth_cache_dir": "/NHNHOME/WORKSPACE/0426030021_A/junhalee/datasets/libero_depth_cache",
15
+ "depth_store": "da3_da3metric-large_r256.zarr",
16
+ "depth_repr": "depth_colormap",
17
+ "depth_cmap": "turbo",
18
+ "depth_power": 0.5,
19
+ "output_dir": "./runs_idm",
20
+ "num_actions": 8,
21
+ "image_size": 256,
22
+ "batch_size": 512,
23
+ "num_workers": 16,
24
+ "color_jitter": false,
25
+ "safe_skip": true,
26
+ "backbone": "facebook/dinov3-vits16-pretrain-lvd1689m",
27
+ "input_mode": "source_depth_flow",
28
+ "freeze_backbone": true,
29
+ "proprio_dim": 8,
30
+ "head_type": "xattn",
31
+ "head_hidden": 1024,
32
+ "head_layers": 3,
33
+ "proprio_hidden": 256,
34
+ "decoder_layers": 4,
35
+ "decoder_heads": 8,
36
+ "dropout": 0.1,
37
+ "huber_delta": 0.1,
38
+ "gripper_weight": 1.0,
39
+ "gripper_idx": 6,
40
+ "action_dim": 7,
41
+ "encoder_input_size": 224,
42
+ "norm_mode": "zscore",
43
+ "norm_stats_path": "fluxflow/norm_stats/libero_norm.json",
44
+ "lr": 0.001,
45
+ "lr_scheduler": "cosine",
46
+ "max_steps": 50000,
47
+ "warmup_steps": 500,
48
+ "save_every": 2000,
49
+ "grad_accum": 4,
50
+ "max_grad_norm": 1.0,
51
+ "mixed_precision": "bf16",
52
+ "seed": 42,
53
+ "wandb_project": "fluxflow-idm",
54
+ "wandb_entity": null,
55
+ "run_name": null,
56
+ "log_every": 20,
57
+ "val_every": 500,
58
+ "num_val_samples": 64,
59
+ "val_fraction": 0.05,
60
+ "split_seed": 0,
61
+ "profile": false,
62
+ "pretrained_model": "black-forest-labs/FLUX.2-klein-base-4B",
63
+ "flux_lora_ckpt": null,
64
+ "mix_flux_p": 0.0,
65
+ "num_inference_steps": 4,
66
+ "flux_guidance_scale": 1.0,
67
+ "flux_batch_size": 0,
68
+ "overfit_episodes": 0,
69
+ "overfit_n": 0,
70
+ "overfit_sample_mode": "linspace",
71
+ "append_progress": true,
72
+ "on_the_fly_flow": false,
73
+ "flow_model": "raft",
74
+ "flow_scale": 0.0,
75
+ "flow_sigma": 0.15,
76
+ "flow_repr": "uv",
77
+ "flow_chunk_size": 32,
78
+ "waft_cfg": "config/a1/tar-c-t.json",
79
+ "waft_ckpt": "/NHNHOME/WORKSPACE/0426030021_A/junhalee/.cache/simvla/waft/checkpoints/waft-a1.pth",
80
+ "waft_dir": "/NHNHOME/WORKSPACE/0426030021_A/junhalee/.cache/simvla/waft",
81
+ "waft_da_ckpt": null,
82
+ "raft_ckpt": "/NHNHOME/WORKSPACE/0426030021_A/junhalee/.cache/simvla/raft/checkpoints/raft-things.pth",
83
+ "raft_dir": "/NHNHOME/WORKSPACE/0426030021_A/junhalee/.cache/simvla/raft",
84
+ "raft_iters": 20,
85
+ "raft_small": false,
86
+ "_overrides": [
87
+ "batch_size",
88
+ "data_dir",
89
+ "grad_accum",
90
+ "max_steps",
91
+ "num_workers",
92
+ "safe_skip"
93
+ ]
94
+ },
95
+ "step": 34000
96
+ }