Robotics
Safetensors
vision-language-action-model
InternVLA-A1-3B / train_config.json
Jia-Zeng's picture
release InternVLA-A1-3B pretrained on InternData-A1 together with agibot-world
6fe7dc4 verified
{
"dataset": {
"type": "qwena1",
"repo_id": "agibotworld/task_480 agibotworld/task_544 agibotworld/task_501 agibotworld/task_372 agibotworld/task_422 agibotworld/task_537 agibotworld/task_366 agibotworld/task_561 agibotworld/task_568 agibotworld/task_421 agibotworld/task_361 agibotworld/task_551 agibotworld/task_363 agibotworld/task_582 agibotworld/task_445 agibotworld/task_444 agibotworld/task_373 agibotworld/task_464 agibotworld/task_542 agibotworld/task_470 agibotworld/task_541 agibotworld/task_487 agibotworld/task_454 agibotworld/task_540 agibotworld/task_431 agibotworld/task_596 agibotworld/task_545 agibotworld/task_362 agibotworld/task_587 agibotworld/task_468 agibotworld/task_616 agibotworld/task_566 agibotworld/task_609 agibotworld/task_365 agibotworld/task_455 agibotworld/task_590 agibotworld/task_683 agibotworld/task_600 agibotworld/task_390 agibotworld/task_613 agibotworld/task_563 agibotworld/task_369 agibotworld/task_567 agibotworld/task_573 agibotworld/task_392 agibotworld/task_368 agibotworld/task_360 agibotworld/task_532 agibotworld/task_351 agibotworld/task_491 agibotworld/task_438 agibotworld/task_357 agibotworld/task_528 agibotworld/task_498 agibotworld/task_511 agibotworld/task_604 agibotworld/task_602 agibotworld/task_429 agibotworld/task_509 agibotworld/task_385 agibotworld/task_521 agibotworld/task_619 agibotworld/task_503 agibotworld/task_424 agibotworld/task_398 agibotworld/task_453 agibotworld/task_374 agibotworld/task_486 agibotworld/task_529 agibotworld/task_520 agibotworld/task_471 agibotworld/task_446 agibotworld/task_465 agibotworld/task_485 agibotworld/task_377 agibotworld/task_512 agibotworld/task_440 agibotworld/task_555 agibotworld/task_688 agibotworld/task_352 agibotworld/task_570 agibotworld/task_507 agibotworld/task_575 agibotworld/task_389 agibotworld/task_494 agibotworld/task_356 agibotworld/task_378 agibotworld/task_558 agibotworld/task_376 agibotworld/task_358 agibotworld/task_506 agibotworld/task_451 agibotworld/task_463 agibotworld/task_550 agibotworld/task_589 agibotworld/task_621 agibotworld/task_466 agibotworld/task_574 agibotworld/task_414 agibotworld/task_504 agibotworld/task_692 agibotworld/task_327 agibotworld/task_588 agibotworld/task_533 agibotworld/task_508 agibotworld/task_525 agibotworld/task_515 agibotworld/task_664 agibotworld/task_375 agibotworld/task_388 agibotworld/task_527 agibotworld/task_359 agibotworld/task_593 agibotworld/task_543 agibotworld/task_483 agibotworld/task_434 agibotworld/task_607 agibotworld/task_425 agibotworld/task_695 agibotworld/task_474 agibotworld/task_689 agibotworld/task_433 agibotworld/task_580 agibotworld/task_410 agibotworld/task_478 agibotworld/task_367 agibotworld/task_597 agibotworld/task_534 agibotworld/task_452 agibotworld/task_522 agibotworld/task_535 agibotworld/task_658 agibotworld/task_462 agibotworld/task_492 agibotworld/task_556 agibotworld/task_603 agibotworld/task_682 agibotworld/task_477 agibotworld/task_599 agibotworld/task_681 agibotworld/task_354 agibotworld/task_524 agibotworld/task_497 agibotworld/task_584 agibotworld/task_598",
"root": null,
"episodes": null,
"image_transforms": {
"enable": false,
"max_num_transforms": 3,
"random_order": false,
"tfs": {
"brightness": {
"weight": 1.0,
"type": "ColorJitter",
"kwargs": {
"brightness": [
0.8,
1.2
]
}
},
"contrast": {
"weight": 1.0,
"type": "ColorJitter",
"kwargs": {
"contrast": [
0.8,
1.2
]
}
},
"saturation": {
"weight": 1.0,
"type": "ColorJitter",
"kwargs": {
"saturation": [
0.5,
1.5
]
}
},
"hue": {
"weight": 1.0,
"type": "ColorJitter",
"kwargs": {
"hue": [
-0.05,
0.05
]
}
},
"sharpness": {
"weight": 1.0,
"type": "SharpnessJitter",
"kwargs": {
"sharpness": [
0.5,
1.5
]
}
},
"affine": {
"weight": 1.0,
"type": "RandomAffine",
"kwargs": {
"degrees": [
-5.0,
5.0
],
"translate": [
0.05,
0.05
]
}
}
}
},
"revision": null,
"use_imagenet_stats": true,
"use_external_stats": true,
"video_backend": "torchcodec",
"streaming": false,
"dist_loading": true,
"buffer_size": 1024,
"action_mode": "delta",
"repack_transforms": {
"inputs": [],
"outputs": []
},
"data_transforms": {
"inputs": [
{
"type": "delta_action",
"mask": null,
"mapping": {}
},
{
"type": "resize_with_pad",
"height": 224,
"width": 224,
"mode": "bilinear"
},
{
"type": "remap_image_key",
"mapping": {}
},
{
"type": "qwena1_processor",
"pretrained_model_name_or_path": "Qwen/Qwen3-VL-2B-Instruct",
"max_length": 48,
"task_key": "task",
"padding_side": "right",
"padding": "max_length",
"truncation": true,
"spatial_merge_size": 2,
"vision_start_token_id": 151652,
"vision_end_token_id": 151653,
"image_token_id": 151655,
"process": null
},
{
"type": "normalize",
"selected_keys": null,
"mode": "mean_std",
"norm_stats": {}
},
{
"type": "compose_fields",
"mapping": {}
},
{
"type": "pad_state_and_action",
"max_state_dim": 32,
"max_action_dim": 32
},
{
"type": "unify_qwena1_inputs"
}
],
"outputs": []
},
"model_transforms": {
"inputs": [],
"outputs": []
},
"height": 224,
"width": 224,
"max_state_dim": 32,
"max_action_dim": 32
},
"env": null,
"policy": {
"type": "qwena1",
"n_obs_steps": 1,
"input_features": {
"observation.state": {
"type": "STATE",
"shape": [
32
]
}
},
"output_features": {
"action": {
"type": "ACTION",
"shape": [
32
]
}
},
"device": "cuda",
"use_amp": false,
"push_to_hub": false,
"repo_id": "jcaiaq/qwena1",
"private": null,
"tags": null,
"license": null,
"pretrained_path": null,
"qwen3_vl_variant": "qwen3_vl_28l",
"action_expert_variant": "qwen3_28l",
"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": 0,
"scheduler_decay_steps": 700000,
"scheduler_decay_lr": 5e-05,
"tokenizer_max_length": 48,
"freeze_vision_encoder": false,
"train_expert_only": false,
"train_vlm_only": false,
"scale_factor": 8,
"lambda_gen": 0.01
},
"output_dir": "/mnt/shared-storage-user/internvla/Users/caijunhao/lerobot/outputs/qwena1/2025_12_19_17_36_47-qwena1-agibotworld-delta-28l-pretrain",
"job_name": "2025_12_19_17_36_47-qwena1-agibotworld-delta-28l-pretrain",
"resume": false,
"seed": 7777,
"num_workers": 8,
"batch_size": 16,
"steps": 700000,
"eval_freq": 20000,
"log_freq": 100,
"save_checkpoint": true,
"save_freq": 10000,
"use_policy_training_preset": true,
"optimizer": {
"type": "adamw",
"lr": 5e-05,
"weight_decay": 0.01,
"grad_clip_norm": 1.0,
"betas": [
0.9,
0.95
],
"eps": 1e-08
},
"scheduler": {
"type": "cosine_decay_with_warmup",
"num_warmup_steps": 0,
"num_decay_steps": 700000,
"peak_lr": 5e-05,
"decay_lr": 5e-05
},
"eval": {
"n_episodes": 50,
"batch_size": 50,
"use_async_envs": false
},
"wandb": {
"enable": true,
"disable_artifact": false,
"project": "lerobot_qwena1",
"entity": null,
"notes": null,
"run_id": "oz1n1l3h",
"mode": "offline"
},
"checkpoint_path": null,
"rename_map": {}
}