Add files using upload-large-folder tool
Browse files- .DS_Store +0 -0
- README.md +52 -0
- speech_tokenizer/hifigan_spiritlm_base/config.json +55 -0
- speech_tokenizer/hifigan_spiritlm_base/generator.pt +3 -0
- speech_tokenizer/hifigan_spiritlm_base/speakers.txt +4 -0
- speech_tokenizer/hifigan_spiritlm_base/styles.txt +34 -0
- speech_tokenizer/hifigan_spiritlm_expressive_w2v2/config.json +60 -0
- speech_tokenizer/hifigan_spiritlm_expressive_w2v2/generator.pt +3 -0
- speech_tokenizer/hifigan_spiritlm_expressive_w2v2/speakers.txt +4 -0
- speech_tokenizer/hubert_25hz/L11_quantizer_500.pt +3 -0
- speech_tokenizer/hubert_25hz/mhubert_base_25hz.pt +3 -0
- speech_tokenizer/style_encoder_w2v2/config.json +321 -0
- speech_tokenizer/style_encoder_w2v2/pytorch_model.bin +3 -0
- speech_tokenizer/vqvae_f0_quantizer/config.yaml +59 -0
- speech_tokenizer/vqvae_f0_quantizer/model.pt +3 -0
- spiritlm_model/spirit-lm-base-7b/config.json +28 -0
- spiritlm_model/spirit-lm-base-7b/generation_config.json +6 -0
- spiritlm_model/spirit-lm-base-7b/pytorch_model.bin +3 -0
- spiritlm_model/spirit-lm-base-7b/special_tokens_map.json +23 -0
- spiritlm_model/spirit-lm-base-7b/tokenizer.model +3 -0
- spiritlm_model/spirit-lm-base-7b/tokenizer_config.json +42 -0
- spiritlm_model/spirit-lm-expressive-7b/config.json +28 -0
- spiritlm_model/spirit-lm-expressive-7b/generation_config.json +6 -0
- spiritlm_model/spirit-lm-expressive-7b/pytorch_model.bin +3 -0
- spiritlm_model/spirit-lm-expressive-7b/special_tokens_map.json +23 -0
- spiritlm_model/spirit-lm-expressive-7b/tokenizer.model +3 -0
- spiritlm_model/spirit-lm-expressive-7b/tokenizer_config.json +42 -0
.DS_Store
ADDED
Binary file (6.15 kB). View file
|
|
README.md
ADDED
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Spirit LM Checkpoints
|
2 |
+
|
3 |
+
## Download Checkpoints
|
4 |
+
To access and download Spirit LM Checkpoints, please request the model artifacts in this link:
|
5 |
+
|
6 |
+
[https://ai.meta.com/resources/models-and-libraries/spirit-lm-downloads/](https://ai.meta.com/resources/models-and-libraries/spirit-lm-downloads/)
|
7 |
+
|
8 |
+
Upon approval, you will then receive an email with download links to each model artifact.
|
9 |
+
|
10 |
+
Please note that Spirit LM is made available under the **FAIR Noncommercial Research License**
|
11 |
+
found in the [LICENSE](../LICENSE) file in the root directory of this source tree and Acceptable Use Policy.
|
12 |
+
|
13 |
+
## Structure
|
14 |
+
The checkpoints directory should look like this:
|
15 |
+
```
|
16 |
+
checkpoints/
|
17 |
+
├── README.md
|
18 |
+
├── speech_tokenizer
|
19 |
+
│ ├── hifigan_spiritlm_base
|
20 |
+
│ │ ├── config.json
|
21 |
+
│ │ ├── generator.pt
|
22 |
+
│ │ ├── speakers.txt
|
23 |
+
│ │ └── styles.txt
|
24 |
+
│ ├── hifigan_spiritlm_expressive_w2v2
|
25 |
+
│ │ ├── config.json
|
26 |
+
│ │ ├── generator.pt
|
27 |
+
│ │ └── speakers.txt
|
28 |
+
│ ├── hubert_25hz
|
29 |
+
│ │ ├── L11_quantizer_500.pt
|
30 |
+
│ │ └── mhubert_base_25hz.pt
|
31 |
+
│ ├── style_encoder_w2v2
|
32 |
+
│ │ ├── config.json
|
33 |
+
│ │ └── pytorch_model.bin
|
34 |
+
│ └── vqvae_f0_quantizer
|
35 |
+
│ ├── config.yaml
|
36 |
+
│ └── model.pt
|
37 |
+
└── spiritlm_model
|
38 |
+
├── spirit-lm-base-7b
|
39 |
+
│ ├── config.json
|
40 |
+
│ ├── generation_config.json
|
41 |
+
│ ├── pytorch_model.bin
|
42 |
+
│ ├── special_tokens_map.json
|
43 |
+
│ ├── tokenizer_config.json
|
44 |
+
│ └── tokenizer.model
|
45 |
+
└── spirit-lm-expressive-7b
|
46 |
+
├── config.json
|
47 |
+
├── generation_config.json
|
48 |
+
├── pytorch_model.bin
|
49 |
+
├── special_tokens_map.json
|
50 |
+
├── tokenizer_config.json
|
51 |
+
└── tokenizer.model
|
52 |
+
```
|
speech_tokenizer/hifigan_spiritlm_base/config.json
ADDED
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
|
3 |
+
"resblock": "1",
|
4 |
+
"num_gpus": 0,
|
5 |
+
"batch_size": 16,
|
6 |
+
"learning_rate": 0.0002,
|
7 |
+
"adam_b1": 0.8,
|
8 |
+
"adam_b2": 0.99,
|
9 |
+
"lr_decay": 0.999,
|
10 |
+
"seed": 1234,
|
11 |
+
|
12 |
+
"upsample_rates": [5,4,4,4,2],
|
13 |
+
"upsample_kernel_sizes": [11,8,8,8,4],
|
14 |
+
"upsample_initial_channel": 512,
|
15 |
+
"resblock_kernel_sizes": [3,7,11],
|
16 |
+
"resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
|
17 |
+
"num_embeddings": 501,
|
18 |
+
"embedding_dim": 128,
|
19 |
+
"model_in_dim": 384,
|
20 |
+
|
21 |
+
"segment_size": 8960,
|
22 |
+
"code_hop_size": 640,
|
23 |
+
"f0": false,
|
24 |
+
"num_mels": 80,
|
25 |
+
"num_freq": 1025,
|
26 |
+
"n_fft": 1024,
|
27 |
+
"hop_size": 256,
|
28 |
+
"win_size": 1024,
|
29 |
+
|
30 |
+
"multispkr": "from_input_file",
|
31 |
+
"num_speakers": 4,
|
32 |
+
"multistyle": "from_input_file",
|
33 |
+
"num_styles": 34,
|
34 |
+
|
35 |
+
"dur_prediction_weight": 1.0,
|
36 |
+
"dur_predictor_params": {
|
37 |
+
"encoder_embed_dim": 128,
|
38 |
+
"var_pred_hidden_dim": 128,
|
39 |
+
"var_pred_kernel_size": 3,
|
40 |
+
"var_pred_dropout": 0.5
|
41 |
+
},
|
42 |
+
|
43 |
+
"sampling_rate": 16000,
|
44 |
+
|
45 |
+
"fmin": 0,
|
46 |
+
"fmax": 8000,
|
47 |
+
"fmax_for_loss": null,
|
48 |
+
|
49 |
+
"num_workers": 4,
|
50 |
+
|
51 |
+
"dist_config": {
|
52 |
+
"dist_backend": "nccl",
|
53 |
+
"dist_url": "env://"
|
54 |
+
}
|
55 |
+
}
|
speech_tokenizer/hifigan_spiritlm_base/generator.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d66c49067aeff93b14b038f143c2dc9ed981671956512d8e702897416f13c459
|
3 |
+
size 57512631
|
speech_tokenizer/hifigan_spiritlm_base/speakers.txt
ADDED
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
1 |
+
ex04
|
2 |
+
ex02
|
3 |
+
ex03
|
4 |
+
ex01
|
speech_tokenizer/hifigan_spiritlm_base/styles.txt
ADDED
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
read-default
|
2 |
+
read-enunciated
|
3 |
+
read-confused
|
4 |
+
read-laughing
|
5 |
+
read-whisper
|
6 |
+
read-sad
|
7 |
+
read-happy
|
8 |
+
conv-projected
|
9 |
+
conv-default
|
10 |
+
conv-sympathetic
|
11 |
+
conv-fast
|
12 |
+
conv-disgusted
|
13 |
+
conv-laughing
|
14 |
+
conv-calm
|
15 |
+
conv-sarcastic
|
16 |
+
conv-whisper
|
17 |
+
conv-angry
|
18 |
+
conv-sad
|
19 |
+
conv-happy
|
20 |
+
conv-enunciated
|
21 |
+
conv-awe
|
22 |
+
read-singing
|
23 |
+
conv-confused
|
24 |
+
conv-fearful
|
25 |
+
conv-narration
|
26 |
+
conv-sleepy
|
27 |
+
conv-child
|
28 |
+
conv-animal
|
29 |
+
conv-childdir
|
30 |
+
conv-animaldir
|
31 |
+
conv-bored
|
32 |
+
conv-desire
|
33 |
+
conv-nonverbal
|
34 |
+
read-narration
|
speech_tokenizer/hifigan_spiritlm_expressive_w2v2/config.json
ADDED
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
|
3 |
+
"resblock": "1",
|
4 |
+
"num_gpus": 0,
|
5 |
+
"batch_size": 128,
|
6 |
+
"learning_rate": 0.0002,
|
7 |
+
"adam_b1": 0.8,
|
8 |
+
"adam_b2": 0.99,
|
9 |
+
"lr_decay": 0.999,
|
10 |
+
"seed": 1234,
|
11 |
+
|
12 |
+
"upsample_rates": [5,4,4,4,2],
|
13 |
+
"upsample_kernel_sizes": [11,8,8,8,4],
|
14 |
+
"upsample_initial_channel": 512,
|
15 |
+
"resblock_kernel_sizes": [3,7,11],
|
16 |
+
"resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
|
17 |
+
|
18 |
+
"multispkr": "from_input_file",
|
19 |
+
"multistyle": null,
|
20 |
+
|
21 |
+
"dur_prediction_weight": 1.0,
|
22 |
+
"dur_predictor_params": {
|
23 |
+
"encoder_embed_dim": 128,
|
24 |
+
"var_pred_hidden_dim": 128,
|
25 |
+
"var_pred_kernel_size": 3,
|
26 |
+
"var_pred_dropout": 0.5
|
27 |
+
},
|
28 |
+
|
29 |
+
"segment_size": 17920,
|
30 |
+
"code_hop_size": 640,
|
31 |
+
"f0_hop_size": 1280,
|
32 |
+
"style_hop_size": 16000,
|
33 |
+
|
34 |
+
"num_embeddings": 501,
|
35 |
+
"num_f0_tokens": 64,
|
36 |
+
"num_style_tokens": 100,
|
37 |
+
"num_speakers": 4,
|
38 |
+
|
39 |
+
"embedding_dim": 128,
|
40 |
+
"model_in_dim": 512,
|
41 |
+
|
42 |
+
"num_mels": 80,
|
43 |
+
"num_freq": 1025,
|
44 |
+
"n_fft": 1024,
|
45 |
+
"hop_size": 256,
|
46 |
+
"win_size": 1024,
|
47 |
+
|
48 |
+
"sampling_rate": 16000,
|
49 |
+
|
50 |
+
"fmin": 0,
|
51 |
+
"fmax": 8000,
|
52 |
+
"fmax_for_loss": null,
|
53 |
+
|
54 |
+
"num_workers": 4,
|
55 |
+
|
56 |
+
"dist_config": {
|
57 |
+
"dist_backend": "nccl",
|
58 |
+
"dist_url": "env://"
|
59 |
+
}
|
60 |
+
}
|
speech_tokenizer/hifigan_spiritlm_expressive_w2v2/generator.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9de97cbc336e6113c27f17560988a22e76df25af6a8e944ad01676aabec31326
|
3 |
+
size 59414584
|
speech_tokenizer/hifigan_spiritlm_expressive_w2v2/speakers.txt
ADDED
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
1 |
+
ex04
|
2 |
+
ex02
|
3 |
+
ex03
|
4 |
+
ex01
|
speech_tokenizer/hubert_25hz/L11_quantizer_500.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:06b408c0a487f0218e8aba52ca30d9de54e0c36af8bebfd151265647d221080b
|
3 |
+
size 5222060
|
speech_tokenizer/hubert_25hz/mhubert_base_25hz.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1421f060cf92b9d2ea72dcecd7f30ce549e1112e24b62b43f2ec3026301051bb
|
3 |
+
size 383333938
|
speech_tokenizer/style_encoder_w2v2/config.json
ADDED
@@ -0,0 +1,321 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "facebook/wav2vec2-base",
|
3 |
+
"activation_dropout": 0.0,
|
4 |
+
"adapter_kernel_size": 3,
|
5 |
+
"adapter_stride": 2,
|
6 |
+
"add_adapter": false,
|
7 |
+
"apply_spec_augment": true,
|
8 |
+
"architectures": [
|
9 |
+
"Wav2Vec2ForPooledSequenceClassification"
|
10 |
+
],
|
11 |
+
"attention_dropout": 0.1,
|
12 |
+
"bos_token_id": 1,
|
13 |
+
"classifier_proj_size": 256,
|
14 |
+
"codevector_dim": 256,
|
15 |
+
"contrastive_logits_temperature": 0.1,
|
16 |
+
"conv_bias": false,
|
17 |
+
"conv_dim": [
|
18 |
+
512,
|
19 |
+
512,
|
20 |
+
512,
|
21 |
+
512,
|
22 |
+
512,
|
23 |
+
512,
|
24 |
+
512
|
25 |
+
],
|
26 |
+
"conv_kernel": [
|
27 |
+
10,
|
28 |
+
3,
|
29 |
+
3,
|
30 |
+
3,
|
31 |
+
3,
|
32 |
+
2,
|
33 |
+
2
|
34 |
+
],
|
35 |
+
"conv_stride": [
|
36 |
+
5,
|
37 |
+
2,
|
38 |
+
2,
|
39 |
+
2,
|
40 |
+
2,
|
41 |
+
2,
|
42 |
+
2
|
43 |
+
],
|
44 |
+
"ctc_loss_reduction": "sum",
|
45 |
+
"ctc_zero_infinity": false,
|
46 |
+
"diversity_loss_weight": 0.1,
|
47 |
+
"do_stable_layer_norm": false,
|
48 |
+
"eos_token_id": 2,
|
49 |
+
"feat_extract_activation": "gelu",
|
50 |
+
"feat_extract_norm": "group",
|
51 |
+
"feat_proj_dropout": 0.1,
|
52 |
+
"feat_quantizer_dropout": 0.0,
|
53 |
+
"final_dropout": 0.0,
|
54 |
+
"freeze_feat_extract_train": true,
|
55 |
+
"hidden_act": "gelu",
|
56 |
+
"hidden_dropout": 0.1,
|
57 |
+
"hidden_size": 768,
|
58 |
+
"id2label": {
|
59 |
+
"0": "0",
|
60 |
+
"1": "1",
|
61 |
+
"10": "10",
|
62 |
+
"11": "11",
|
63 |
+
"12": "12",
|
64 |
+
"13": "13",
|
65 |
+
"14": "14",
|
66 |
+
"15": "15",
|
67 |
+
"16": "16",
|
68 |
+
"17": "17",
|
69 |
+
"18": "18",
|
70 |
+
"19": "19",
|
71 |
+
"2": "2",
|
72 |
+
"20": "20",
|
73 |
+
"21": "21",
|
74 |
+
"22": "22",
|
75 |
+
"23": "23",
|
76 |
+
"24": "24",
|
77 |
+
"25": "25",
|
78 |
+
"26": "26",
|
79 |
+
"27": "27",
|
80 |
+
"28": "28",
|
81 |
+
"29": "29",
|
82 |
+
"3": "3",
|
83 |
+
"30": "30",
|
84 |
+
"31": "31",
|
85 |
+
"32": "32",
|
86 |
+
"33": "33",
|
87 |
+
"34": "34",
|
88 |
+
"35": "35",
|
89 |
+
"36": "36",
|
90 |
+
"37": "37",
|
91 |
+
"38": "38",
|
92 |
+
"39": "39",
|
93 |
+
"4": "4",
|
94 |
+
"40": "40",
|
95 |
+
"41": "41",
|
96 |
+
"42": "42",
|
97 |
+
"43": "43",
|
98 |
+
"44": "44",
|
99 |
+
"45": "45",
|
100 |
+
"46": "46",
|
101 |
+
"47": "47",
|
102 |
+
"48": "48",
|
103 |
+
"49": "49",
|
104 |
+
"5": "5",
|
105 |
+
"50": "50",
|
106 |
+
"51": "51",
|
107 |
+
"52": "52",
|
108 |
+
"53": "53",
|
109 |
+
"54": "54",
|
110 |
+
"55": "55",
|
111 |
+
"56": "56",
|
112 |
+
"57": "57",
|
113 |
+
"58": "58",
|
114 |
+
"59": "59",
|
115 |
+
"6": "6",
|
116 |
+
"60": "60",
|
117 |
+
"61": "61",
|
118 |
+
"62": "62",
|
119 |
+
"63": "63",
|
120 |
+
"64": "64",
|
121 |
+
"65": "65",
|
122 |
+
"66": "66",
|
123 |
+
"67": "67",
|
124 |
+
"68": "68",
|
125 |
+
"69": "69",
|
126 |
+
"7": "7",
|
127 |
+
"70": "70",
|
128 |
+
"71": "71",
|
129 |
+
"72": "72",
|
130 |
+
"73": "73",
|
131 |
+
"74": "74",
|
132 |
+
"75": "75",
|
133 |
+
"76": "76",
|
134 |
+
"77": "77",
|
135 |
+
"78": "78",
|
136 |
+
"79": "79",
|
137 |
+
"8": "8",
|
138 |
+
"80": "80",
|
139 |
+
"81": "81",
|
140 |
+
"82": "82",
|
141 |
+
"83": "83",
|
142 |
+
"84": "84",
|
143 |
+
"85": "85",
|
144 |
+
"86": "86",
|
145 |
+
"87": "87",
|
146 |
+
"88": "88",
|
147 |
+
"89": "89",
|
148 |
+
"9": "9",
|
149 |
+
"90": "90",
|
150 |
+
"91": "91",
|
151 |
+
"92": "92",
|
152 |
+
"93": "93",
|
153 |
+
"94": "94",
|
154 |
+
"95": "95",
|
155 |
+
"96": "96",
|
156 |
+
"97": "97",
|
157 |
+
"98": "98",
|
158 |
+
"99": "99"
|
159 |
+
},
|
160 |
+
"initializer_range": 0.02,
|
161 |
+
"intermediate_size": 3072,
|
162 |
+
"label2id": {
|
163 |
+
"0": "0",
|
164 |
+
"1": "1",
|
165 |
+
"10": "10",
|
166 |
+
"11": "11",
|
167 |
+
"12": "12",
|
168 |
+
"13": "13",
|
169 |
+
"14": "14",
|
170 |
+
"15": "15",
|
171 |
+
"16": "16",
|
172 |
+
"17": "17",
|
173 |
+
"18": "18",
|
174 |
+
"19": "19",
|
175 |
+
"2": "2",
|
176 |
+
"20": "20",
|
177 |
+
"21": "21",
|
178 |
+
"22": "22",
|
179 |
+
"23": "23",
|
180 |
+
"24": "24",
|
181 |
+
"25": "25",
|
182 |
+
"26": "26",
|
183 |
+
"27": "27",
|
184 |
+
"28": "28",
|
185 |
+
"29": "29",
|
186 |
+
"3": "3",
|
187 |
+
"30": "30",
|
188 |
+
"31": "31",
|
189 |
+
"32": "32",
|
190 |
+
"33": "33",
|
191 |
+
"34": "34",
|
192 |
+
"35": "35",
|
193 |
+
"36": "36",
|
194 |
+
"37": "37",
|
195 |
+
"38": "38",
|
196 |
+
"39": "39",
|
197 |
+
"4": "4",
|
198 |
+
"40": "40",
|
199 |
+
"41": "41",
|
200 |
+
"42": "42",
|
201 |
+
"43": "43",
|
202 |
+
"44": "44",
|
203 |
+
"45": "45",
|
204 |
+
"46": "46",
|
205 |
+
"47": "47",
|
206 |
+
"48": "48",
|
207 |
+
"49": "49",
|
208 |
+
"5": "5",
|
209 |
+
"50": "50",
|
210 |
+
"51": "51",
|
211 |
+
"52": "52",
|
212 |
+
"53": "53",
|
213 |
+
"54": "54",
|
214 |
+
"55": "55",
|
215 |
+
"56": "56",
|
216 |
+
"57": "57",
|
217 |
+
"58": "58",
|
218 |
+
"59": "59",
|
219 |
+
"6": "6",
|
220 |
+
"60": "60",
|
221 |
+
"61": "61",
|
222 |
+
"62": "62",
|
223 |
+
"63": "63",
|
224 |
+
"64": "64",
|
225 |
+
"65": "65",
|
226 |
+
"66": "66",
|
227 |
+
"67": "67",
|
228 |
+
"68": "68",
|
229 |
+
"69": "69",
|
230 |
+
"7": "7",
|
231 |
+
"70": "70",
|
232 |
+
"71": "71",
|
233 |
+
"72": "72",
|
234 |
+
"73": "73",
|
235 |
+
"74": "74",
|
236 |
+
"75": "75",
|
237 |
+
"76": "76",
|
238 |
+
"77": "77",
|
239 |
+
"78": "78",
|
240 |
+
"79": "79",
|
241 |
+
"8": "8",
|
242 |
+
"80": "80",
|
243 |
+
"81": "81",
|
244 |
+
"82": "82",
|
245 |
+
"83": "83",
|
246 |
+
"84": "84",
|
247 |
+
"85": "85",
|
248 |
+
"86": "86",
|
249 |
+
"87": "87",
|
250 |
+
"88": "88",
|
251 |
+
"89": "89",
|
252 |
+
"9": "9",
|
253 |
+
"90": "90",
|
254 |
+
"91": "91",
|
255 |
+
"92": "92",
|
256 |
+
"93": "93",
|
257 |
+
"94": "94",
|
258 |
+
"95": "95",
|
259 |
+
"96": "96",
|
260 |
+
"97": "97",
|
261 |
+
"98": "98",
|
262 |
+
"99": "99"
|
263 |
+
},
|
264 |
+
"layer_norm_eps": 1e-05,
|
265 |
+
"layerdrop": 0.0,
|
266 |
+
"mask_channel_length": 10,
|
267 |
+
"mask_channel_min_space": 1,
|
268 |
+
"mask_channel_other": 0.0,
|
269 |
+
"mask_channel_prob": 0.0,
|
270 |
+
"mask_channel_selection": "static",
|
271 |
+
"mask_feature_length": 10,
|
272 |
+
"mask_feature_min_masks": 0,
|
273 |
+
"mask_feature_prob": 0.0,
|
274 |
+
"mask_time_length": 10,
|
275 |
+
"mask_time_min_masks": 2,
|
276 |
+
"mask_time_min_space": 1,
|
277 |
+
"mask_time_other": 0.0,
|
278 |
+
"mask_time_prob": 0.05,
|
279 |
+
"mask_time_selection": "static",
|
280 |
+
"model_type": "wav2vec2",
|
281 |
+
"no_mask_channel_overlap": false,
|
282 |
+
"no_mask_time_overlap": false,
|
283 |
+
"num_adapter_layers": 3,
|
284 |
+
"num_attention_heads": 12,
|
285 |
+
"num_codevector_groups": 2,
|
286 |
+
"num_codevectors_per_group": 320,
|
287 |
+
"num_conv_pos_embedding_groups": 16,
|
288 |
+
"num_conv_pos_embeddings": 128,
|
289 |
+
"num_feat_extract_layers": 7,
|
290 |
+
"num_hidden_layers": 12,
|
291 |
+
"num_negatives": 100,
|
292 |
+
"output_hidden_size": 768,
|
293 |
+
"pad_token_id": 0,
|
294 |
+
"proj_codevector_dim": 256,
|
295 |
+
"tdnn_dilation": [
|
296 |
+
1,
|
297 |
+
2,
|
298 |
+
3,
|
299 |
+
1,
|
300 |
+
1
|
301 |
+
],
|
302 |
+
"tdnn_dim": [
|
303 |
+
512,
|
304 |
+
512,
|
305 |
+
512,
|
306 |
+
512,
|
307 |
+
1500
|
308 |
+
],
|
309 |
+
"tdnn_kernel": [
|
310 |
+
5,
|
311 |
+
3,
|
312 |
+
3,
|
313 |
+
1,
|
314 |
+
1
|
315 |
+
],
|
316 |
+
"torch_dtype": "float32",
|
317 |
+
"transformers_version": "4.25.1",
|
318 |
+
"use_weighted_layer_sum": false,
|
319 |
+
"vocab_size": 32,
|
320 |
+
"xvector_output_dim": 512
|
321 |
+
}
|
speech_tokenizer/style_encoder_w2v2/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:819b7a87bfcf4041252c3f1825915924e10d5a06a2a66da546beaca98c7ab8bc
|
3 |
+
size 378451177
|
speech_tokenizer/vqvae_f0_quantizer/config.yaml
ADDED
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
seed: 1234
|
3 |
+
|
4 |
+
# Data
|
5 |
+
f0_path: ''
|
6 |
+
p_train: 0.95
|
7 |
+
min_frames: null
|
8 |
+
batch_size: 128
|
9 |
+
features: f0_interp,vuv
|
10 |
+
out_features: norm_f0_interp,vuv
|
11 |
+
segment_size: null
|
12 |
+
segment_multi: 16
|
13 |
+
num_workers: 4
|
14 |
+
vuv_scale: 2
|
15 |
+
speaker_stats: ''
|
16 |
+
recon_loss_fn: l1_loss
|
17 |
+
|
18 |
+
|
19 |
+
# Optimization
|
20 |
+
learning_rate: 0.0002
|
21 |
+
adam_b1: 0.8
|
22 |
+
adam_b2: 0.99
|
23 |
+
lr_decay: 0.999
|
24 |
+
lambda_commit: 0.02
|
25 |
+
|
26 |
+
# VQ params
|
27 |
+
vq_params:
|
28 |
+
l_bins: 64
|
29 |
+
emb_width: 128
|
30 |
+
mu: 0.99
|
31 |
+
levels: 1
|
32 |
+
|
33 |
+
# Encoder params
|
34 |
+
encoder_params:
|
35 |
+
input_emb_width: 2
|
36 |
+
output_emb_width: 128
|
37 |
+
levels: 1
|
38 |
+
downs_t:
|
39 |
+
- 4
|
40 |
+
strides_t:
|
41 |
+
- 2
|
42 |
+
width: 32
|
43 |
+
depth: 4
|
44 |
+
m_conv: 1.0
|
45 |
+
dilation_growth_rate: 3
|
46 |
+
|
47 |
+
# Decoder params
|
48 |
+
decoder_params:
|
49 |
+
input_emb_width: 2
|
50 |
+
output_emb_width: 128
|
51 |
+
levels: 1
|
52 |
+
downs_t:
|
53 |
+
- 4
|
54 |
+
strides_t:
|
55 |
+
- 2
|
56 |
+
width: 32
|
57 |
+
depth: 4
|
58 |
+
m_conv: 1.0
|
59 |
+
dilation_growth_rate: 3
|
speech_tokenizer/vqvae_f0_quantizer/model.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f4321b0c19b47279ab3d76c4b6e85bbc439156a4dd6478919856accaf4180382
|
3 |
+
size 2600601
|
spiritlm_model/spirit-lm-base-7b/config.json
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"LlamaForCausalLM"
|
4 |
+
],
|
5 |
+
"attention_bias": false,
|
6 |
+
"attention_dropout": 0.0,
|
7 |
+
"bos_token_id": 1,
|
8 |
+
"eos_token_id": 2,
|
9 |
+
"hidden_act": "silu",
|
10 |
+
"hidden_size": 4096,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 11008,
|
13 |
+
"max_position_embeddings": 16384,
|
14 |
+
"mlp_bias": false,
|
15 |
+
"model_type": "llama",
|
16 |
+
"num_attention_heads": 32,
|
17 |
+
"num_hidden_layers": 32,
|
18 |
+
"num_key_value_heads": 32,
|
19 |
+
"pretraining_tp": 1,
|
20 |
+
"rms_norm_eps": 1e-05,
|
21 |
+
"rope_scaling": null,
|
22 |
+
"rope_theta": 100000.0,
|
23 |
+
"tie_word_embeddings": false,
|
24 |
+
"torch_dtype": "bfloat16",
|
25 |
+
"transformers_version": "4.42.4",
|
26 |
+
"use_cache": true,
|
27 |
+
"vocab_size": 32512
|
28 |
+
}
|
spiritlm_model/spirit-lm-base-7b/generation_config.json
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 1,
|
4 |
+
"eos_token_id": 2,
|
5 |
+
"transformers_version": "4.42.4"
|
6 |
+
}
|
spiritlm_model/spirit-lm-base-7b/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:403d6afbb451adb7df7f33a19914256f1ec5a02ec73889b5a86cca03521b70eb
|
3 |
+
size 13485316654
|
spiritlm_model/spirit-lm-base-7b/special_tokens_map.json
ADDED
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": {
|
3 |
+
"content": "<s>",
|
4 |
+
"lstrip": false,
|
5 |
+
"normalized": false,
|
6 |
+
"rstrip": false,
|
7 |
+
"single_word": false
|
8 |
+
},
|
9 |
+
"eos_token": {
|
10 |
+
"content": "</s>",
|
11 |
+
"lstrip": false,
|
12 |
+
"normalized": false,
|
13 |
+
"rstrip": false,
|
14 |
+
"single_word": false
|
15 |
+
},
|
16 |
+
"unk_token": {
|
17 |
+
"content": "<unk>",
|
18 |
+
"lstrip": false,
|
19 |
+
"normalized": false,
|
20 |
+
"rstrip": false,
|
21 |
+
"single_word": false
|
22 |
+
}
|
23 |
+
}
|
spiritlm_model/spirit-lm-base-7b/tokenizer.model
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0e0f741d0605911d0793969b84ac89cebb00e3b7b1a575713e40079410df1ff9
|
3 |
+
size 508928
|
spiritlm_model/spirit-lm-base-7b/tokenizer_config.json
ADDED
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_bos_token": true,
|
3 |
+
"add_eos_token": false,
|
4 |
+
"add_prefix_space": true,
|
5 |
+
"added_tokens_decoder": {
|
6 |
+
"0": {
|
7 |
+
"content": "<unk>",
|
8 |
+
"lstrip": false,
|
9 |
+
"normalized": false,
|
10 |
+
"rstrip": false,
|
11 |
+
"single_word": false,
|
12 |
+
"special": true
|
13 |
+
},
|
14 |
+
"1": {
|
15 |
+
"content": "<s>",
|
16 |
+
"lstrip": false,
|
17 |
+
"normalized": false,
|
18 |
+
"rstrip": false,
|
19 |
+
"single_word": false,
|
20 |
+
"special": true
|
21 |
+
},
|
22 |
+
"2": {
|
23 |
+
"content": "</s>",
|
24 |
+
"lstrip": false,
|
25 |
+
"normalized": false,
|
26 |
+
"rstrip": false,
|
27 |
+
"single_word": false,
|
28 |
+
"special": true
|
29 |
+
}
|
30 |
+
},
|
31 |
+
"bos_token": "<s>",
|
32 |
+
"clean_up_tokenization_spaces": false,
|
33 |
+
"eos_token": "</s>",
|
34 |
+
"legacy": true,
|
35 |
+
"model_max_length": 1000000000000000019884624838656,
|
36 |
+
"pad_token": null,
|
37 |
+
"sp_model_kwargs": {},
|
38 |
+
"spaces_between_special_tokens": false,
|
39 |
+
"tokenizer_class": "LlamaTokenizer",
|
40 |
+
"unk_token": "<unk>",
|
41 |
+
"use_default_system_prompt": false
|
42 |
+
}
|
spiritlm_model/spirit-lm-expressive-7b/config.json
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"LlamaForCausalLM"
|
4 |
+
],
|
5 |
+
"attention_bias": false,
|
6 |
+
"attention_dropout": 0.0,
|
7 |
+
"bos_token_id": 1,
|
8 |
+
"eos_token_id": 2,
|
9 |
+
"hidden_act": "silu",
|
10 |
+
"hidden_size": 4096,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 11008,
|
13 |
+
"max_position_embeddings": 16384,
|
14 |
+
"mlp_bias": false,
|
15 |
+
"model_type": "llama",
|
16 |
+
"num_attention_heads": 32,
|
17 |
+
"num_hidden_layers": 32,
|
18 |
+
"num_key_value_heads": 32,
|
19 |
+
"pretraining_tp": 1,
|
20 |
+
"rms_norm_eps": 1e-05,
|
21 |
+
"rope_scaling": null,
|
22 |
+
"rope_theta": 100000.0,
|
23 |
+
"tie_word_embeddings": false,
|
24 |
+
"torch_dtype": "bfloat16",
|
25 |
+
"transformers_version": "4.42.4",
|
26 |
+
"use_cache": true,
|
27 |
+
"vocab_size": 32768
|
28 |
+
}
|
spiritlm_model/spirit-lm-expressive-7b/generation_config.json
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 1,
|
4 |
+
"eos_token_id": 2,
|
5 |
+
"transformers_version": "4.42.4"
|
6 |
+
}
|
spiritlm_model/spirit-lm-expressive-7b/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8effa9480ca74af529dd3244eab4820c2b1bdfbd3196e571676258525b1f3e05
|
3 |
+
size 13489510958
|
spiritlm_model/spirit-lm-expressive-7b/special_tokens_map.json
ADDED
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": {
|
3 |
+
"content": "<s>",
|
4 |
+
"lstrip": false,
|
5 |
+
"normalized": false,
|
6 |
+
"rstrip": false,
|
7 |
+
"single_word": false
|
8 |
+
},
|
9 |
+
"eos_token": {
|
10 |
+
"content": "</s>",
|
11 |
+
"lstrip": false,
|
12 |
+
"normalized": false,
|
13 |
+
"rstrip": false,
|
14 |
+
"single_word": false
|
15 |
+
},
|
16 |
+
"unk_token": {
|
17 |
+
"content": "<unk>",
|
18 |
+
"lstrip": false,
|
19 |
+
"normalized": false,
|
20 |
+
"rstrip": false,
|
21 |
+
"single_word": false
|
22 |
+
}
|
23 |
+
}
|
spiritlm_model/spirit-lm-expressive-7b/tokenizer.model
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1608b90876103c6c3c67ca079ab9d3c5ee4e7707acf869103e728a4d30626643
|
3 |
+
size 514364
|
spiritlm_model/spirit-lm-expressive-7b/tokenizer_config.json
ADDED
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_bos_token": true,
|
3 |
+
"add_eos_token": false,
|
4 |
+
"add_prefix_space": true,
|
5 |
+
"added_tokens_decoder": {
|
6 |
+
"0": {
|
7 |
+
"content": "<unk>",
|
8 |
+
"lstrip": false,
|
9 |
+
"normalized": false,
|
10 |
+
"rstrip": false,
|
11 |
+
"single_word": false,
|
12 |
+
"special": true
|
13 |
+
},
|
14 |
+
"1": {
|
15 |
+
"content": "<s>",
|
16 |
+
"lstrip": false,
|
17 |
+
"normalized": false,
|
18 |
+
"rstrip": false,
|
19 |
+
"single_word": false,
|
20 |
+
"special": true
|
21 |
+
},
|
22 |
+
"2": {
|
23 |
+
"content": "</s>",
|
24 |
+
"lstrip": false,
|
25 |
+
"normalized": false,
|
26 |
+
"rstrip": false,
|
27 |
+
"single_word": false,
|
28 |
+
"special": true
|
29 |
+
}
|
30 |
+
},
|
31 |
+
"bos_token": "<s>",
|
32 |
+
"clean_up_tokenization_spaces": false,
|
33 |
+
"eos_token": "</s>",
|
34 |
+
"legacy": true,
|
35 |
+
"model_max_length": 1000000000000000019884624838656,
|
36 |
+
"pad_token": null,
|
37 |
+
"sp_model_kwargs": {},
|
38 |
+
"spaces_between_special_tokens": false,
|
39 |
+
"tokenizer_class": "LlamaTokenizer",
|
40 |
+
"unk_token": "<unk>",
|
41 |
+
"use_default_system_prompt": false
|
42 |
+
}
|