evo1_so100_vla / config.json
Beilinghamburger's picture
Upload policy weights, train config and readme
58c6120 verified
{
"type": "evo1",
"n_obs_steps": 1,
"normalization_mapping": {
"VISUAL": "IDENTITY",
"STATE": "MEAN_STD",
"ACTION": "MEAN_STD"
},
"input_features": {
"observation.state": {
"type": "STATE",
"shape": [
6
]
},
"observation.images.front": {
"type": "VISUAL",
"shape": [
3,
480,
640
]
},
"observation.images.wrist": {
"type": "VISUAL",
"shape": [
3,
480,
640
]
}
},
"output_features": {
"action": {
"type": "ACTION",
"shape": [
6
]
}
},
"device": "cuda",
"use_amp": false,
"push_to_hub": true,
"repo_id": "Beilinghamburger/evo1_so100_vla",
"private": null,
"tags": null,
"license": null,
"chunk_size": 50,
"n_action_steps": 50,
"max_state_dim": 24,
"max_action_dim": 24,
"resize_imgs_with_padding": [
448,
448
],
"empty_cameras": 0,
"vlm_model_name": "OpenGVLab/InternVL3-1B",
"load_vlm_weights": true,
"add_image_special_tokens": false,
"attention_mode": "cross_attn",
"prefix_length": 0,
"pad_language_to": "max_length",
"action_head": "flowmatching",
"dropout": 0.2,
"num_layers": 8,
"per_action_dim": 24,
"state_dim": 24,
"action_dim": 1200,
"use_augmentation": true,
"binarize_gripper": false,
"freeze_vision_encoder": false,
"train_expert_only": false,
"train_state_proj": true,
"optimizer_lr": 1e-05,
"optimizer_betas": [
0.9,
0.95
],
"optimizer_eps": 1e-08,
"optimizer_weight_decay": 0.001,
"optimizer_grad_clip_norm": 1.0,
"scheduler_warmup_steps": 1000,
"scheduler_decay_steps": 30000,
"scheduler_decay_lr": 2.5e-06,
"batch_size": 16,
"max_steps": 30000,
"log_interval": 10,
"ckpt_interval": 2500,
"save_dir": "./checkpoints/evo1",
"resume": true,
"resume_path": "/home/usyd/Evo-1-e/so100_evo1/checkpoint",
"resume_pretrain": true,
"num_workers": 4,
"disable_wandb": true,
"return_cls_only": false,
"run_name": "Evo1_Default",
"dataset_type": "lerobot",
"dataset_config_path": "",
"data_paths": null,
"num_steps": 10,
"min_period": 0.004,
"max_period": 4.0
}