Hugging Face's logo

zaleni
/
WSA-Base

Robotics
LeRobot
Safetensors
English
embodied-ai
robot-learning
vision-language-action
world-model
Model card Files
xet
 Community
1
WSA-Base
File size: 2,677 Bytes
3f027de
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
{
    "type": "TBot_SA1",
    "n_obs_steps": 1,
    "input_features": {
        "observation.state": {
            "type": "STATE",
            "shape": [
                32
            ]
        }
    },
    "output_features": {
        "action": {
            "type": "ACTION",
            "shape": [
                32
            ]
        }
    },
    "device": "cuda:0",
    "use_amp": false,
    "push_to_hub": false,
    "repo_id": "zaleni/TBot-SA1-Base",
    "private": null,
    "tags": null,
    "license": null,
    "pretrained_path": null,
    "qwen3_vl_variant": "qwen3_vl_28l",
    "action_expert_variant": "qwen3_28l",
    "qwen3_vl_pretrained_path": "Qwen/Qwen3-VL-2B-Instruct",
    "dtype": "bfloat16",
    "chunk_size": 50,
    "n_action_steps": 50,
    "max_state_dim": 32,
    "max_action_dim": 32,
    "num_inference_steps": 10,
    "time_sampling_beta_alpha": 1.5,
    "time_sampling_beta_beta": 1.0,
    "time_sampling_scale": 0.999,
    "time_sampling_offset": 0.001,
    "min_period": 0.004,
    "max_period": 4.0,
    "image_resolution": [
        224,
        224
    ],
    "empty_cameras": 0,
    "normalization_mapping": {
        "VISUAL": "IDENTITY",
        "STATE": "IDENTITY",
        "ACTION": "IDENTITY"
    },
    "gradient_checkpointing": false,
    "compile_model": false,
    "compile_mode": "max-autotune",
    "optimizer_lr": 5e-05,
    "optimizer_betas": [
        0.9,
        0.95
    ],
    "optimizer_eps": 1e-08,
    "optimizer_weight_decay": 0.01,
    "optimizer_grad_clip_norm": 1.0,
    "scheduler_warmup_steps": 2000,
    "scheduler_decay_steps": 300000,
    "scheduler_decay_lr": 1e-05,
    "tokenizer_max_length": 48,
    "freeze_vision_encoder": false,
    "train_expert_only": false,
    "train_vlm_only": false,
    "scale_factor": 8,
    "lambda_gen": 0.01,
    "cosmos_tokenizer_path_or_name": "nvidia/Cosmos-Tokenizer-CI8x8",
    "enable_3d_queries": true,
    "num_3d_query_tokens": 432,
    "da3_alignment_mode": "query_decoder",
    "da3_query_resampler_layers": 1,
    "da3_query_resampler_ff_mult": 1,
    "query_layer_indices": [
        13,
        19,
        23,
        27
    ],
    "da3_variant": "auto",
    "da3_teacher_layers": [
        11,
        15,
        19,
        23
    ],
    "da3_query_dim": 2048,
    "da3_tokens_per_view": 1296,
    "da3_num_views": 3,
    "lambda_3d": 0.01,
    "da3_model_path_or_name": "depth-anything/DA3-LARGE-1.1",
    "da3_model_name": null,
    "da3_code_root": null,
    "da3_teacher_process_res": 504,
    "da3_layer_weights": [
        1.0,
        1.2,
        1.4,
        1.6
    ],
    "future_query_init_std": 0.02,
    "log_da3_teacher_timing": true
}