hand_roboState_v2 / config.json
cheng000's picture
Upload OpenVLAForActionPrediction
85d254b verified
{
"_name_or_path": "/home/robo/Projects/p7_vla/openvla_hand_bin_model_roboState_v2/runs/openvla-7b+adapthand_dataset+b8+lr-2e-05+lora-r32+dropout-0.0",
"arch_specifier": "no-align+fused-gelu-mlp",
"architectures": [
"OpenVLAForActionPrediction"
],
"auto_map": {
"AutoConfig": "configuration_prismatic.OpenVLAConfig",
"AutoModelForVision2Seq": "modeling_prismatic.OpenVLAForActionPrediction"
},
"hf_llm_id": "meta-llama/Llama-2-7b-hf",
"image_resize_strategy": "resize-naive",
"image_sizes": [
224,
224
],
"llm_backbone_id": "llama2-7b-pure",
"llm_max_length": 2048,
"model_type": "openvla",
"n_action_bins": 256,
"norm_stats": {
"adapthand_dataset": {
"action": {
"mask": [
true,
true,
true,
true,
true,
true,
false
],
"max": [
0.0934932678937912,
0.07073106616735458,
0.13955646753311157,
25.86539649963379,
28.09152603149414,
22.63760757446289,
1.0
],
"mean": [
-0.0004674541705753654,
-7.01767532973463e-07,
-0.0026117712259292603,
-0.9660950303077698,
0.07697256654500961,
0.23577749729156494,
0.5337038040161133
],
"min": [
-0.11865614354610443,
-0.06380447000265121,
-0.12263807654380798,
-18.1720027923584,
-11.647391319274902,
-20.10640525817871,
-0.0
],
"q01": [
-0.06869871757924557,
-0.04083644054830075,
-0.09486489862203598,
-12.397282724380494,
-6.393810477256775,
-10.171583023071289,
-0.0
],
"q99": [
0.06211262740194787,
0.04004575975239276,
0.12924241289496421,
14.330417251586896,
6.437622961997966,
9.655190410613951,
1.0
],
"std": [
0.024569833651185036,
0.015697475522756577,
0.04801836237311363,
5.1880879402160645,
2.133551836013794,
3.278003454208374,
0.33055388927459717
]
},
"num_trajectories": 20,
"num_transitions": 1864,
"proprio": {
"max": [
0.7340260148048401,
0.3007831573486328,
0.40622445940971375,
90.72176361083984,
2.3114571571350098,
-115.23106384277344,
0.0,
1.0
],
"mean": [
0.5701256990432739,
0.20672091841697693,
0.2367880940437317,
73.28887939453125,
-14.52593994140625,
-134.03717041015625,
0.0,
0.5337038040161133
],
"min": [
0.41738706827163696,
0.1260789930820465,
0.12270695716142654,
53.13225555419922,
-23.411903381347656,
-156.28704833984375,
0.0,
-0.0
],
"q01": [
0.4208172786235809,
0.13053419023752214,
0.12882332623004913,
56.18941757202148,
-22.627037773132326,
-153.44824035644533,
0.0,
-0.0
],
"q99": [
0.7245610493421555,
0.29450163006782526,
0.39202539771795275,
89.14522323608398,
-0.17190513432029214,
-117.6465699005127,
0.0,
1.0
],
"std": [
0.06350336223840714,
0.042166419327259064,
0.06861510872840881,
7.5594868659973145,
4.653266906738281,
8.585131645202637,
0.0,
0.33055388927459717
]
}
}
},
"output_projector_states": false,
"pad_to_multiple_of": 64,
"pad_token_id": 32000,
"text_config": {
"model_type": "llama",
"pad_token_id": 32000,
"torch_dtype": "bfloat16",
"vocab_size": 32064
},
"timm_model_ids": [
"vit_large_patch14_reg4_dinov2.lvd142m",
"vit_so400m_patch14_siglip_224"
],
"timm_override_act_layers": [
null,
null
],
"torch_dtype": "bfloat16",
"transformers_version": "4.40.1",
"use_fused_vision_backbone": true,
"vision_backbone_id": "dinosiglip-vit-so-224px"
}