{ "type": "tadpole", "n_obs_steps": 1, "input_features": { "observation.images.top": { "type": "VISUAL", "shape": [ 3, 512, 512 ] }, "observation.images.side": { "type": "VISUAL", "shape": [ 3, 512, 512 ] }, "observation.images.iso": { "type": "VISUAL", "shape": [ 3, 512, 512 ] }, "observation.state": { "type": "STATE", "shape": [ 7 ] } }, "output_features": { "action": { "type": "ACTION", "shape": [ 7 ] } }, "device": "cpu", "use_amp": false, "use_peft": false, "push_to_hub": true, "repo_id": null, "private": null, "tags": null, "license": null, "pretrained_path": null, "pretrained_revision": null, "chunk_size": 8, "n_action_steps": 8, "normalization_mapping": { "VISUAL": "IDENTITY", "STATE": "IDENTITY", "ACTION": "IDENTITY" }, "image_size": 512, "dim": 256, "layers": 6, "heads": 8, "grid": 16, "redraw_patch": 8, "tokenizer_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct", "tokenizer_max_length": 48, "vocab_size": 49280, "token_ids": [ 2, 260, 272, 284, 288, 543, 614, 826, 1478, 3115, 3154, 3921, 4231, 4870, 5222, 5927, 7842, 9571, 11986, 12536, 13917, 14297, 14362, 14973, 17201, 17967, 19609, 24036, 27865, 28948, 31413, 32577 ], "step_scale": [ 0.04, 0.04, 0.04, 0.04, 0.04, 0.04, 0.125 ], "joint_limits": [ [ 0.1, 2.5132741228718345 ], [ 1.5707963267948966, 4.71238898038469 ], [ -2.199114857512855, 2.199114857512855 ], [ -1.8849555921538759, 1.8849555921538759 ], [ -3.141592653589793, 3.141592653589793 ], [ -3.141592653589793, 3.141592653589793 ] ], "state_dropout": 0.5, "redraw_weight": 1.0, "optimizer_lr": 0.0003, "optimizer_weight_decay": 1e-10, "optimizer_betas": [ 0.9, 0.95 ] }