Hugging Face's logo

allenai
/
MolmoAct-7B-D-Pretrain-0812

Robotics
Transformers
Safetensors
English
molmoact
image-text-to-text
molmo
olmo
reasoning
vla
manipulation
custom_code
Model card Files
xet
Community
1
MolmoAct-7B-D-Pretrain-0812 / config.json
hqfang's picture
hqfang
update weights from bf16 to fp32
8fd5f92
raw
history blame contribute delete
9.72 kB
 {
"adapter_config": {
"attention_dropout": 0.0,
"float32_attention": true,
"head_dim": 72,
"hidden_act": "silu",
"hidden_size": 1152,
"image_feature_dropout": 0.0,
"image_padding_embed": null,
"initializer_range": 0.02,
"intermediate_size": 18944,
"model_type": "",
"num_attention_heads": 16,
"num_key_value_heads": 16,
"residual_dropout": 0.0,
"text_hidden_size": 3584,
"vit_layers": [
-3,
-9
]
},
"architectures": [
"MolmoActForActionReasoning"
],
"auto_map": {
"AutoConfig": "configuration_molmoact.MolmoActConfig",
"AutoModelForImageTextToText": "modeling_molmoact.MolmoActForActionReasoning"
},
"image_patch_id": 152066,
"initializer_range": 0.02,
"llm_config": {
"additional_vocab_size": 128,
"attention_dropout": 0.0,
"embedding_dropout": 0.0,
"head_dim": 128,
"hidden_act": "silu",
"hidden_size": 3584,
"initializer_range": 0.02,
"intermediate_size": 18944,
"layer_norm_eps": 1e-06,
"max_position_embeddings": 4096,
"model_type": "molmoact_llm",
"norm_after": false,
"num_attention_heads": 28,
"num_hidden_layers": 28,
"num_key_value_heads": 4,
"qk_norm_type": "olmo",
"qkv_bias": true,
"residual_dropout": 0.0,
"rope_scaling": null,
"rope_theta": 1000000.0,
"use_cache": true,
"use_qk_norm": false,
"vocab_size": 152064
},
"model_type": "molmoact",
"n_action_bins": 256,
"norm_stats": {
"bc_z": {
"action": {
"mask": [
true,
true,
true,
true,
true,
true,
false
],
"max": [
0.2165454924106598,
0.1251407265663147,
0.10772687941789627,
0.33544227480888367,
0.28117990493774414,
0.40614867210388184,
1.0
],
"mean": [
-0.009958467446267605,
0.0008958321413956583,
0.004995597992092371,
0.00029755113064311445,
-0.008735382929444313,
-0.030693737789988518,
0.8344562649726868
],
"min": [
-0.1677047461271286,
-0.14630407094955444,
-0.10066790133714676,
-0.29421567916870117,
-0.32101404666900635,
-0.4635624885559082,
0.0
],
"q01": [
-0.09220654994249344,
-0.06456145539879798,
-0.049121275544166565,
-0.11594625547528267,
-0.14152548640966414,
-0.2251061636209488,
0.0
],
"q99": [
0.07628866866230968,
0.058019736707210584,
0.052540797740221024,
0.11740604028105736,
0.11703975558280955,
0.16729306846857078,
1.0
],
"std": [
0.03053455986082554,
0.0231423731893301,
0.020641816779971123,
0.04155943542718887,
0.046427831053733826,
0.0769818127155304,
0.3610210120677948
]
},
"num_trajectories": 43264,
"num_transitions": 6015535,
"proprio": {
"max": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"mean": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"min": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"q01": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"q99": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"std": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
]
}
},
"bridge_orig": {
"action": {
"mask": [
true,
true,
true,
true,
true,
true,
false
],
"max": [
0.41691166162490845,
0.25864794850349426,
0.21218234300613403,
3.122201919555664,
1.8618112802505493,
6.280478477478027,
1.0
],
"mean": [
0.0002334194869035855,
0.00013004911306779832,
-0.00012762474943883717,
-0.0001556558854645118,
-0.0004039328487124294,
0.00023557482927571982,
0.5764579176902771
],
"min": [
-0.4007510244846344,
-0.13874775171279907,
-0.22553899884223938,
-3.2010786533355713,
-1.8618112802505493,
-6.279075622558594,
0.0
],
"q01": [
-0.02872725307941437,
-0.04170349963009357,
-0.026093858778476715,
-0.08092105075716972,
-0.09288699507713317,
-0.20718276381492615,
0.0
],
"q99": [
0.028309678435325586,
0.040855254605412394,
0.040161586627364146,
0.08192047759890528,
0.07792850524187081,
0.20382574498653397,
1.0
],
"std": [
0.009765930473804474,
0.013689135201275349,
0.012667362578213215,
0.028534092009067535,
0.030637972056865692,
0.07691419124603271,
0.4973701536655426
]
},
"num_trajectories": 60064,
"num_transitions": 2135463,
"proprio": {
"max": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"mean": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"min": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"q01": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"q99": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"std": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
]
}
},
"fractal20220817_data": {
"action": {
"mask": [
true,
true,
true,
true,
true,
true,
false
],
"max": [
2.9984593391418457,
22.09052848815918,
2.7507524490356445,
1.570636510848999,
1.5321086645126343,
1.5691522359848022,
1.0
],
"mean": [
0.006987582892179489,
0.006265917327255011,
-0.01262515690177679,
0.04333311319351196,
-0.005756212864071131,
0.0009130256366916001,
0.5354204773902893
],
"min": [
-2.0204520225524902,
-5.497899532318115,
-2.031663417816162,
-1.569917917251587,
-1.569892168045044,
-1.570419430732727,
0.0
],
"q01": [
-0.22453527510166169,
-0.14820013284683228,
-0.231589707583189,
-0.3517994859814644,
-0.4193011274933815,
-0.43643461108207704,
0.0
],
"q99": [
0.17824687153100965,
0.14938379630446405,
0.21842354819178575,
0.5892666035890578,
0.35272657424211445,
0.44796681255102094,
1.0
],
"std": [
0.0692116990685463,
0.05970962345600128,
0.07353084534406662,
0.15610496699810028,
0.13164450228214264,
0.14593800902366638,
0.497110515832901
]
},
"num_trajectories": 87212,
"num_transitions": 3786400,
"proprio": {
"max": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"mean": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"min": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"q01": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"q99": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"std": [
0.0,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
]
}
}
},
"tie_word_embeddings": false,
"torch_dtype": "float32",
"transformers_version": "4.52.3",
"use_cache": true,
"vit_config": {
"attention_dropout": 0.0,
"float32_attention": true,
"head_dim": 72,
"hidden_act": "gelu_pytorch_tanh",
"hidden_size": 1152,
"image_default_input_size": [
378,
378
],
"image_num_pos": 729,
"image_patch_size": 14,
"initializer_range": 0.02,
"intermediate_size": 4304,
"layer_norm_eps": 1e-06,
"model_type": "molmoact_vit",
"num_attention_heads": 16,
"num_hidden_layers": 27,
"num_key_value_heads": 16,
"patch_bias": true,
"pre_layernorm": false,
"residual_dropout": 0.0,
"use_cls_token": false
}
}