pinecoresystems commited on
Commit
060027c
·
verified ·
1 Parent(s): c0b215f

Mirror config.json

Browse files
Files changed (1) hide show
  1. config.json +274 -0
config.json ADDED
@@ -0,0 +1,274 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "MossAudioTokenizerModel"
4
+ ],
5
+ "auto_map": {
6
+ "AutoConfig": "configuration_moss_audio_tokenizer.MossAudioTokenizerConfig",
7
+ "AutoModel": "modeling_moss_audio_tokenizer.MossAudioTokenizerModel"
8
+ },
9
+ "causal_transformer_context_duration": 10,
10
+ "code_dim": 768,
11
+ "decoder_kwargs": [
12
+ {
13
+ "causal": true,
14
+ "conv_layout": true,
15
+ "d_model": 1280,
16
+ "dim_feedforward": 5120,
17
+ "gating": "none",
18
+ "input_dimension": 768,
19
+ "layer_scale": 0.01,
20
+ "max_period": 10000,
21
+ "module_type": "Transformer",
22
+ "norm": "layer_norm",
23
+ "num_heads": 20,
24
+ "num_layers": 32,
25
+ "output_dimension": 1280,
26
+ "positional_embedding": "rope"
27
+ },
28
+ {
29
+ "module_type": "PatchedPretransform",
30
+ "patch_size": 2
31
+ },
32
+ {
33
+ "causal": true,
34
+ "conv_layout": true,
35
+ "d_model": 768,
36
+ "dim_feedforward": 3072,
37
+ "gating": "none",
38
+ "input_dimension": 640,
39
+ "layer_scale": 0.01,
40
+ "max_period": 10000,
41
+ "module_type": "Transformer",
42
+ "norm": "layer_norm",
43
+ "num_heads": 12,
44
+ "num_layers": 12,
45
+ "output_dimension": 768,
46
+ "positional_embedding": "rope"
47
+ },
48
+ {
49
+ "module_type": "PatchedPretransform",
50
+ "patch_size": 2
51
+ },
52
+ {
53
+ "causal": true,
54
+ "conv_layout": true,
55
+ "d_model": 768,
56
+ "dim_feedforward": 3072,
57
+ "gating": "none",
58
+ "input_dimension": 384,
59
+ "layer_scale": 0.01,
60
+ "max_period": 10000,
61
+ "module_type": "Transformer",
62
+ "norm": "layer_norm",
63
+ "num_heads": 12,
64
+ "num_layers": 12,
65
+ "output_dimension": 768,
66
+ "positional_embedding": "rope"
67
+ },
68
+ {
69
+ "module_type": "PatchedPretransform",
70
+ "patch_size": 2
71
+ },
72
+ {
73
+ "causal": true,
74
+ "conv_layout": true,
75
+ "d_model": 768,
76
+ "dim_feedforward": 3072,
77
+ "gating": "none",
78
+ "input_dimension": 384,
79
+ "layer_scale": 0.01,
80
+ "max_period": 10000,
81
+ "module_type": "Transformer",
82
+ "norm": "layer_norm",
83
+ "num_heads": 12,
84
+ "num_layers": 12,
85
+ "output_dimension": 240,
86
+ "positional_embedding": "rope"
87
+ },
88
+ {
89
+ "module_type": "PatchedPretransform",
90
+ "patch_size": 240
91
+ }
92
+ ],
93
+ "downsample_rate": 1920,
94
+ "dtype": "float32",
95
+ "encoder_kwargs": [
96
+ {
97
+ "module_type": "PatchedPretransform",
98
+ "patch_size": 240
99
+ },
100
+ {
101
+ "causal": true,
102
+ "conv_layout": true,
103
+ "d_model": 768,
104
+ "dim_feedforward": 3072,
105
+ "gating": "none",
106
+ "input_dimension": 240,
107
+ "layer_scale": 0.01,
108
+ "max_period": 10000,
109
+ "module_type": "Transformer",
110
+ "norm": "layer_norm",
111
+ "num_heads": 12,
112
+ "num_layers": 12,
113
+ "output_dimension": 384,
114
+ "positional_embedding": "rope"
115
+ },
116
+ {
117
+ "module_type": "PatchedPretransform",
118
+ "patch_size": 2
119
+ },
120
+ {
121
+ "causal": true,
122
+ "conv_layout": true,
123
+ "d_model": 768,
124
+ "dim_feedforward": 3072,
125
+ "gating": "none",
126
+ "input_dimension": 768,
127
+ "layer_scale": 0.01,
128
+ "max_period": 10000,
129
+ "module_type": "Transformer",
130
+ "norm": "layer_norm",
131
+ "num_heads": 12,
132
+ "num_layers": 12,
133
+ "output_dimension": 384,
134
+ "positional_embedding": "rope"
135
+ },
136
+ {
137
+ "module_type": "PatchedPretransform",
138
+ "patch_size": 2
139
+ },
140
+ {
141
+ "causal": true,
142
+ "conv_layout": true,
143
+ "d_model": 768,
144
+ "dim_feedforward": 3072,
145
+ "gating": "none",
146
+ "input_dimension": 768,
147
+ "layer_scale": 0.01,
148
+ "max_period": 10000,
149
+ "module_type": "Transformer",
150
+ "norm": "layer_norm",
151
+ "num_heads": 12,
152
+ "num_layers": 12,
153
+ "output_dimension": 640,
154
+ "positional_embedding": "rope"
155
+ },
156
+ {
157
+ "module_type": "PatchedPretransform",
158
+ "patch_size": 2
159
+ },
160
+ {
161
+ "causal": true,
162
+ "conv_layout": true,
163
+ "d_model": 1280,
164
+ "dim_feedforward": 5120,
165
+ "gating": "none",
166
+ "input_dimension": 1280,
167
+ "layer_scale": 0.01,
168
+ "max_period": 10000,
169
+ "module_type": "Transformer",
170
+ "norm": "layer_norm",
171
+ "num_heads": 20,
172
+ "num_layers": 32,
173
+ "output_dimension": 768,
174
+ "positional_embedding": "rope"
175
+ }
176
+ ],
177
+ "model_type": "moss-audio-tokenizer",
178
+ "quantizer_kwargs": {
179
+ "codebook_dim": 8,
180
+ "codebook_size": 1024,
181
+ "input_dim": 768,
182
+ "num_quantizers": 32,
183
+ "output_dim": 768,
184
+ "quantizer_type": "rlfq",
185
+ "rvq_dim": 512
186
+ },
187
+ "quantizer_type": "rlfq",
188
+ "reversed_decoder_kwargs": [
189
+ {
190
+ "module_type": "PatchedPretransform",
191
+ "patch_size": 240
192
+ },
193
+ {
194
+ "causal": true,
195
+ "conv_layout": true,
196
+ "d_model": 768,
197
+ "dim_feedforward": 3072,
198
+ "gating": "none",
199
+ "input_dimension": 240,
200
+ "layer_scale": 0.01,
201
+ "max_period": 10000,
202
+ "module_type": "Transformer",
203
+ "norm": "layer_norm",
204
+ "num_heads": 12,
205
+ "num_layers": 12,
206
+ "output_dimension": 384,
207
+ "positional_embedding": "rope"
208
+ },
209
+ {
210
+ "module_type": "PatchedPretransform",
211
+ "patch_size": 2
212
+ },
213
+ {
214
+ "causal": true,
215
+ "conv_layout": true,
216
+ "d_model": 768,
217
+ "dim_feedforward": 3072,
218
+ "gating": "none",
219
+ "input_dimension": 768,
220
+ "layer_scale": 0.01,
221
+ "max_period": 10000,
222
+ "module_type": "Transformer",
223
+ "norm": "layer_norm",
224
+ "num_heads": 12,
225
+ "num_layers": 12,
226
+ "output_dimension": 384,
227
+ "positional_embedding": "rope"
228
+ },
229
+ {
230
+ "module_type": "PatchedPretransform",
231
+ "patch_size": 2
232
+ },
233
+ {
234
+ "causal": true,
235
+ "conv_layout": true,
236
+ "d_model": 768,
237
+ "dim_feedforward": 3072,
238
+ "gating": "none",
239
+ "input_dimension": 768,
240
+ "layer_scale": 0.01,
241
+ "max_period": 10000,
242
+ "module_type": "Transformer",
243
+ "norm": "layer_norm",
244
+ "num_heads": 12,
245
+ "num_layers": 12,
246
+ "output_dimension": 640,
247
+ "positional_embedding": "rope"
248
+ },
249
+ {
250
+ "module_type": "PatchedPretransform",
251
+ "patch_size": 2
252
+ },
253
+ {
254
+ "causal": true,
255
+ "conv_layout": true,
256
+ "d_model": 1280,
257
+ "dim_feedforward": 5120,
258
+ "gating": "none",
259
+ "input_dimension": 1280,
260
+ "layer_scale": 0.01,
261
+ "max_period": 10000,
262
+ "module_type": "Transformer",
263
+ "norm": "layer_norm",
264
+ "num_heads": 20,
265
+ "num_layers": 32,
266
+ "output_dimension": 768,
267
+ "positional_embedding": "rope"
268
+ }
269
+ ],
270
+ "sample_rate": 24000,
271
+ "sampling_rate": 24000,
272
+ "transformers_version": "4.56.0.dev0",
273
+ "version": "4.26.1.a"
274
+ }