{ "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": [ "SPRVLAForActionReasoning" ], "auto_map": { "AutoConfig": "configuration_sprvla.SPRVLAConfig", "AutoModelForImageTextToText": "modeling_sprvla.SPRVLAForActionReasoning" }, "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": "sprvla_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": "sprvla", "n_action_bins": 256, "norm_stats": { "sprvla": { "action": { "max": [ 0.06042003631591797, 0.09417290985584259, 0.07019275426864624, 0.2616892158985138, 0.11751057207584381, 0.16968433558940887, 1.0 ], "mean": [ 0.0005706787342205644, 0.0002448957529850304, -3.5987635783385485e-05, 0.00021597897284664214, -0.0004896928439848125, -0.000241481073317118, 0.5570635199546814 ], "min": [ -0.07434078305959702, -0.07339745759963989, -0.06539416313171387, -0.1688285619020462, -0.10289879888296127, -0.2667275667190552, 0.0 ], "q01": [ -0.01538565568625927, -0.021047022193670273, -0.01688069850206375, -0.044314172118902206, -0.03890235349535942, -0.04788423702120781, 0.0 ], "q99": [ 0.014661382883787155, 0.026515591889619827, 0.021398313343524933, 0.04216696694493294, 0.03401297703385353, 0.04957397282123566, 1.0 ], "std": [ 0.005207270849496126, 0.007506529800593853, 0.006415561307221651, 0.013248044066131115, 0.010928540490567684, 0.014873150736093521, 0.49715080857276917 ] }, "num_entries": 1560068 }, "libero_object_no_noops_modified": { "action": { "max": [ 0.9375, 0.8919642567634583, 0.9375, 0.17678570747375488, 0.35035714507102966, 0.1810714304447174, 1.0 ], "mean": [ 0.07096529006958008, 0.13498851656913757, -0.04601382836699486, 0.00123520044144243, 0.006998839322477579, -0.015027612447738647, 0.46428999304771423 ], "min": [ -0.8839285969734192, -0.9375, -0.9375, -0.15000000596046448, -0.29035714268684387, -0.32892856001853943, 0.0 ], "q01": [ -0.5383928418159485, -0.8758928775787354, -0.9375, -0.06964285671710968, -0.11678571254014969, -0.15964286029338837, 0.0 ], "q99": [ 0.8464285731315613, 0.84375, 0.9375, 0.08142857253551483, 0.14892856776714325, 0.0867857113480568, 1.0 ], "std": [ 0.2681235373020172, 0.43846824765205383, 0.4474974274635315, 0.024446550756692886, 0.049355510622262955, 0.042107198387384415, 0.49879148602485657 ] }, "num_trajectories": 454, "num_transitions": 66984, "proprio": { "max": [ 0.14580604434013367, 0.33216384053230286, 0.3857804834842682, 3.4003844261169434, 0.7954911589622498, 0.6642207503318787, 0.0, 0.04104341194033623, -0.00018117300351150334 ], "mean": [ -0.02999030612409115, -0.007947085425257683, 0.20293472707271576, 3.1086409091949463, -0.21404768526554108, -0.11307074874639511, 0.0, 0.029380427673459053, -0.030556727200746536 ], "min": [ -0.1765444278717041, -0.29457300901412964, 0.008128180168569088, 2.2890501022338867, -1.883241891860962, -1.0600427389144897, 0.0, 0.0006495157140307128, -0.041782498359680176 ], "q01": [ -0.14911890715360643, -0.25978428691625594, 0.009925739830359817, 2.7545341420173646, -1.3996034812927245, -0.6867720144987106, 0.0, 0.008197814421728254, -0.04015838988125324 ], "q99": [ 0.09063626825809479, 0.29066365867853167, 0.3370887073874472, 3.2611824750900267, 0.32092821151018125, 0.4037663781642913, 0.0, 0.039891827926039694, -0.009106044843792932 ], "std": [ 0.06694897264242172, 0.17608462274074554, 0.07807064801454544, 0.0868484303355217, 0.33540457487106323, 0.20728276669979095, 0.0, 0.00956575945019722, 0.009197483770549297 ] } } }, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.52.1", "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": "sprvla_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 } }