Hugging Face's logo

drbaph
/
LongCat-AudioDiT-3.5B-bf16

Text-to-Speech
Transformers
Safetensors
Chinese
English
audiodit
comfyui
Model card Files
xet
 Community
drbaph commited on
Commit
0693bcc
·
verified ·
1 Parent(s): 664bf7d

Upload 2 files

Browse files
Files changed (2) hide show
  1. config.json +106 -0
  2. model.safetensors +3 -0
config.json ADDED
@@ -0,0 +1,106 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dit_adaln_type": "global",
3
+ "dit_adaln_use_text_cond": true,
4
+ "dit_bias": true,
5
+ "dit_cross_attn": true,
6
+ "dit_cross_attn_norm": false,
7
+ "dit_depth": 32,
8
+ "dit_dim": 2560,
9
+ "dit_dropout": 0.0,
10
+ "dit_eps": 1e-06,
11
+ "dit_ff_mult": 3.6,
12
+ "dit_heads": 32,
13
+ "dit_long_skip": true,
14
+ "dit_qk_norm": true,
15
+ "dit_text_conv": true,
16
+ "dit_text_dim": 768,
17
+ "dit_use_latent_condition": true,
18
+ "latent_dim": 64,
19
+ "latent_hop": 2048,
20
+ "max_wav_duration": 60,
21
+ "model_type": "audiodit",
22
+ "repa_dit_layer": 8,
23
+ "sampling_rate": 24000,
24
+ "sigma": 0.0,
25
+ "text_add_embed": true,
26
+ "text_encoder_config": {
27
+ "_name_or_path": "ArthurZ/umt5-base",
28
+ "architectures": [
29
+ "UMT5ForConditionalGeneration"
30
+ ],
31
+ "chunk_size_feed_forward": 0,
32
+ "classifier_dropout": 0.0,
33
+ "d_ff": 2048,
34
+ "d_kv": 64,
35
+ "d_model": 768,
36
+ "decoder_start_token_id": 0,
37
+ "dense_act_fn": "gelu_new",
38
+ "dropout_rate": 0.1,
39
+ "dtype": "float32",
40
+ "eos_token_id": 1,
41
+ "feed_forward_proj": "gated-gelu",
42
+ "id2label": {
43
+ "0": "LABEL_0",
44
+ "1": "LABEL_1"
45
+ },
46
+ "initializer_factor": 1.0,
47
+ "is_decoder": false,
48
+ "is_encoder_decoder": true,
49
+ "is_gated_act": true,
50
+ "label2id": {
51
+ "LABEL_0": 0,
52
+ "LABEL_1": 1
53
+ },
54
+ "layer_norm_epsilon": 1e-06,
55
+ "model_type": "umt5",
56
+ "num_decoder_layers": 12,
57
+ "num_heads": 12,
58
+ "num_layers": 12,
59
+ "output_attentions": false,
60
+ "output_hidden_states": false,
61
+ "output_past": true,
62
+ "pad_token_id": 0,
63
+ "problem_type": null,
64
+ "relative_attention_max_distance": 128,
65
+ "relative_attention_num_buckets": 32,
66
+ "return_dict": true,
67
+ "scalable_attention": true,
68
+ "tie_word_embeddings": true,
69
+ "tokenizer_class": "T5Tokenizer",
70
+ "use_cache": true,
71
+ "vocab_size": 256384
72
+ },
73
+ "text_encoder_model": "google/umt5-base",
74
+ "text_norm_feat": true,
75
+ "transformers_version": "5.3.0",
76
+ "vae_config": {
77
+ "c_mults": [
78
+ 1,
79
+ 2,
80
+ 4,
81
+ 8,
82
+ 16
83
+ ],
84
+ "channels": 128,
85
+ "downsample_shortcut": "averaging",
86
+ "downsampling_ratio": 2048,
87
+ "encoder_latent_dim": 128,
88
+ "final_tanh": false,
89
+ "in_channels": 1,
90
+ "in_shortcut": "duplicating",
91
+ "latent_dim": 64,
92
+ "model_type": "audiodit_vae",
93
+ "out_shortcut": "averaging",
94
+ "sample_rate": 24000,
95
+ "scale": 0.71,
96
+ "strides": [
97
+ 2,
98
+ 4,
99
+ 4,
100
+ 8,
101
+ 8
102
+ ],
103
+ "upsample_shortcut": "duplicating",
104
+ "use_snake": true
105
+ }
106
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dd3beb8dd0b7c3e9b221168bcd92497cfcafe5e546ca68501dfc51f5d012f947
3
+ size 7668131810