Hugging Face's logo

bezzam
/
xcodec-hubert-librispeech

Feature Extraction
Transformers
Safetensors
xcodec
Model card Files Community
bezzam HF Staff commited on
Commit
ead4046
·
verified ·
1 Parent(s): a2349ba

Upload model

Browse files
Files changed (2) hide show
  1. config.json +135 -0
  2. model.safetensors +3 -0
config.json ADDED
@@ -0,0 +1,135 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "acoustic_model_config": {
3
+ "codebook_dim": 8,
4
+ "codebook_loss_weight": 1.0,
5
+ "codebook_size": 1024,
6
+ "commitment_loss_weight": 0.25,
7
+ "decoder_hidden_size": 1024,
8
+ "downsampling_ratios": [
9
+ 8,
10
+ 5,
11
+ 4,
12
+ 2
13
+ ],
14
+ "encoder_hidden_size": 64,
15
+ "hidden_size": 256,
16
+ "hop_length": 320,
17
+ "model_type": "dac",
18
+ "n_codebooks": 9,
19
+ "quantizer_dropout": 0,
20
+ "sampling_rate": 16000,
21
+ "upsampling_ratios": [
22
+ 8,
23
+ 5,
24
+ 4,
25
+ 2
26
+ ]
27
+ },
28
+ "architectures": [
29
+ "XcodecModel"
30
+ ],
31
+ "block_dilations": [
32
+ 1,
33
+ 1
34
+ ],
35
+ "channel_ratios": [
36
+ 1,
37
+ 1
38
+ ],
39
+ "codebook_dim": 1024,
40
+ "codebook_size": 1024,
41
+ "decoder_channels": 768,
42
+ "encoder_channels": 768,
43
+ "initializer_range": 0.02,
44
+ "input_channels": 768,
45
+ "kernel_size": 3,
46
+ "model_type": "xcodec",
47
+ "output_channels": 768,
48
+ "sample_rate": 16000,
49
+ "semantic_model_config": {
50
+ "_name_or_path": "facebook/hubert-base-ls960",
51
+ "activation_dropout": 0.1,
52
+ "apply_spec_augment": true,
53
+ "architectures": [
54
+ "HubertModel"
55
+ ],
56
+ "attention_dropout": 0.1,
57
+ "classifier_proj_size": 256,
58
+ "conv_bias": false,
59
+ "conv_dim": [
60
+ 512,
61
+ 512,
62
+ 512,
63
+ 512,
64
+ 512,
65
+ 512,
66
+ 512
67
+ ],
68
+ "conv_kernel": [
69
+ 10,
70
+ 3,
71
+ 3,
72
+ 3,
73
+ 3,
74
+ 2,
75
+ 2
76
+ ],
77
+ "conv_pos_batch_norm": false,
78
+ "conv_stride": [
79
+ 5,
80
+ 2,
81
+ 2,
82
+ 2,
83
+ 2,
84
+ 2,
85
+ 2
86
+ ],
87
+ "ctc_loss_reduction": "sum",
88
+ "ctc_zero_infinity": false,
89
+ "do_stable_layer_norm": false,
90
+ "feat_extract_activation": "gelu",
91
+ "feat_extract_dropout": 0.0,
92
+ "feat_extract_norm": "group",
93
+ "feat_proj_dropout": 0.1,
94
+ "feat_proj_layer_norm": true,
95
+ "final_dropout": 0.1,
96
+ "gradient_checkpointing": false,
97
+ "hidden_act": "gelu",
98
+ "hidden_dropout": 0.1,
99
+ "hidden_dropout_prob": 0.1,
100
+ "hidden_size": 768,
101
+ "initializer_range": 0.02,
102
+ "intermediate_size": 3072,
103
+ "layer_norm_eps": 1e-05,
104
+ "layerdrop": 0.1,
105
+ "mask_feature_length": 10,
106
+ "mask_feature_min_masks": 0,
107
+ "mask_feature_prob": 0.0,
108
+ "mask_time_length": 10,
109
+ "mask_time_min_masks": 2,
110
+ "mask_time_prob": 0.05,
111
+ "model_type": "hubert",
112
+ "num_attention_heads": 12,
113
+ "num_conv_pos_embedding_groups": 16,
114
+ "num_conv_pos_embeddings": 128,
115
+ "num_feat_extract_layers": 7,
116
+ "num_hidden_layers": 12,
117
+ "tokenizer_class": "Wav2Vec2CTCTokenizer",
118
+ "use_weighted_layer_sum": false,
119
+ "vocab_size": 32
120
+ },
121
+ "strides": [
122
+ 1,
123
+ 1
124
+ ],
125
+ "target_bandwidths": [
126
+ 0.5,
127
+ 1,
128
+ 1.5,
129
+ 2,
130
+ 4
131
+ ],
132
+ "torch_dtype": "float32",
133
+ "transformers_version": "4.56.0.dev0",
134
+ "unit_kernel_size": 3
135
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6191971294c216bf6fb7f38fb39e69267a2e3eea6b5469a0952806923b901ce5
3
+ size 709991220