pyp1 commited on
Commit
d31203f
1 Parent(s): 547e4a6

Push model using huggingface_hub.

Browse files
Files changed (2) hide show
  1. config.json +70 -0
  2. model.safetensors +3 -0
config.json ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "seed": 1,
3
+ "precision": "float16",
4
+ "num_workers": 8,
5
+ "resume": false,
6
+ "tb_write_every_n_steps": 10,
7
+ "print_every_n_steps": 400,
8
+ "val_every_n_steps": 1600,
9
+ "lr": 0.03,
10
+ "batch_size": 100,
11
+ "weight_decay": 0.01,
12
+ "warmup_fraction": 0.01,
13
+ "num_epochs": 10,
14
+ "num_steps": 50000,
15
+ "gradient_accumulation_steps": 8,
16
+ "gradient_clip_val": 1.0,
17
+ "early_stop_step": 3200,
18
+ "early_stop_threshold": -1.0,
19
+ "exp_dir": "/data/scratch/pyp/exp_pyp/VoiceCraft/gigaspeech/e5",
20
+ "dataset": "gigaspeech",
21
+ "dataset_dir": "/data/scratch/pyp/datasets/gigaspeech_phn_enc_manifest/xl",
22
+ "pseudo_epoch_size": 3000,
23
+ "phn_folder_name": "phonemes",
24
+ "encodec_folder_name": "encodec_16khz_4codebooks",
25
+ "manifest_name": "manifest_large16khz_lessambi",
26
+ "pad_x": 0,
27
+ "max_num_tokens": 100000,
28
+ "val_max_num_tokens": 6000,
29
+ "num_buckets": 6,
30
+ "dynamic_batching": 1,
31
+ "audio_max_length": 20.0,
32
+ "audio_min_length": 2.0,
33
+ "text_max_length": 400,
34
+ "text_min_length": 10.0,
35
+ "encodec_sr": 50,
36
+ "mask_len_min": 1,
37
+ "mask_len_max": 600,
38
+ "drop_long": 1,
39
+ "eos": 2051,
40
+ "reduced_eog": 1,
41
+ "special_first": 0,
42
+ "n_special": 4,
43
+ "codebook_weight": "[5,1,0.5,0.1]",
44
+ "empty_token": 2048,
45
+ "optimizer_name": "ScaledAdam",
46
+ "reduce_lr_start_step": 3000,
47
+ "reduce_lr_start_epoch": 4,
48
+ "clipping_update_period": 1000,
49
+ "max_mask_portion": 0.9,
50
+ "max_n_spans": 3,
51
+ "shuffle_mask_embedding": 0,
52
+ "mask_sample_dist": "poisson1",
53
+ "min_gap": 5,
54
+ "n_codebooks": 4,
55
+ "text_vocab_size": 100,
56
+ "text_pad_token": 100,
57
+ "audio_vocab_size": 2048,
58
+ "eog": 2049,
59
+ "audio_pad_token": 2050,
60
+ "d_model": 2048,
61
+ "audio_embedding_dim": 2048,
62
+ "text_embedding_dropout": 0.1,
63
+ "audio_embedding_dropout": 0,
64
+ "text_positional_embedding_dropout": 0.1,
65
+ "audio_positional_embedding_dropout": 0.1,
66
+ "trm_dropout": 0.1,
67
+ "nhead": 16,
68
+ "num_decoder_layers": 8,
69
+ "load_model_from": null
70
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6579ad8332d9ae123f7ee93bc373f0f56e67c5ab54e1ed2b4493db8a18059f93
3
+ size 1746809632