{ "format_version": 1, "architecture": "MultiDimWav2Vec2ForPreTraining", "model_config": { "activation_dropout": 0.0, "apply_spec_augment": true, "attention_dropout": 0.0, "codevector_dim": 256, "contrastive_logits_temperature": 0.1, "conv_bias": true, "conv_dim": [ 256, 256 ], "conv_embed": true, "conv_kernel": [ 3, 3 ], "conv_stride": [ 2, 2 ], "diversity_loss_weight": 0.1, "do_stable_layer_norm": true, "feat_extract_norm": "layer", "feat_proj_dropout": 0.0, "feat_quantizer_dropout": 0.0, "hidden_dropout": 0.0, "hidden_size": 240, "initializer_range": 0.02, "input_dim": 3, "intermediate_size": 3072, "layer_norm_eps": 1e-05, "layerdrop": 0.0, "mask_feature_length": 10, "mask_feature_min_masks": 0, "mask_feature_prob": 0.0, "mask_time_length": 10, "mask_time_min_masks": 2, "mask_time_prob": 0.65, "num_attention_heads": 12, "num_codevector_groups": 2, "num_codevectors_per_group": 320, "num_conv_pos_embedding_groups": 16, "num_conv_pos_embeddings": 128, "num_feat_extract_layers": 2, "num_hidden_layers": 6, "num_negatives": 100, "output_attentions": false, "output_hidden_size": 768, "output_hidden_states": false, "pad_token_id": 0, "proj_codevector_dim": 256, "rotary_pos_embed": false, "scale_logits_in_quantization": true, "use_rms_norm": false }, "preprocessing": { "sampling_rate_hz": 100, "component_order": "ZNE", "tensor_layout": "batch, component, sample", "dtype": "float32", "reference_window_samples": 3001, "normalization": "per-component demean, divide by population std + 1e-10", "zero_std": "replace zero std by 1 before division", "missing_components": "Legacy training randomly copies a present component into missing channels. Supply complete components for deterministic inference; no filling is performed by the loader or encode()." }, "source_checkpoint_sha256": "6e97c53c562e9ba6bc96df85d8d208b89ff84e80a25468825f29bb5bace5eacb" }