dramabox-tts
lucasnewman commited on
Commit
b1c4944
·
verified ·
1 Parent(s): f3de920

Upload 3 files

Browse files
config.json ADDED
@@ -0,0 +1,47 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "dramabox-tts",
3
+ "architecture": "DiT-FlowMatching",
4
+ "base_model": "ltx-2.3-22b-dev-audio-only",
5
+ "parameters": "3.3B",
6
+ "num_layers": 48,
7
+ "audio_inner_dim": 2048,
8
+ "audio_num_attention_heads": 32,
9
+ "audio_attention_head_dim": 64,
10
+ "audio_cross_attention_dim": 2048,
11
+ "denoising_steps": 30,
12
+ "scheduler": "euler_flow_matching",
13
+ "text_encoder": "mlx-community/gemma-3-12b-it-8bit",
14
+ "text_encoder_hidden_size": 3840,
15
+ "ic_lora": {
16
+ "rank": 128,
17
+ "alpha": 128,
18
+ "merged": true,
19
+ "training_version": "v13",
20
+ "text_dropout": 0.4,
21
+ "training_steps": "v12@3000 + v13@1000"
22
+ },
23
+ "audio": {
24
+ "sample_rate": 48000,
25
+ "vae_channels": 8,
26
+ "mel_bins": 16,
27
+ "fps": 25.0
28
+ },
29
+ "inference_defaults": {
30
+ "cfg_scale": 2.5,
31
+ "stg_scale": 1.5,
32
+ "rescale_scale": "auto",
33
+ "modality_scale": 1.0,
34
+ "duration_multiplier": 1.1,
35
+ "seed": 42
36
+ },
37
+ "files": {
38
+ "transformer": "dramabox-dit-v1.safetensors",
39
+ "audio_components": "dramabox-audio-components.safetensors",
40
+ "silence_latent": "assets/silence_latent_frame.pt"
41
+ },
42
+ "mlx_audio": {
43
+ "source_repo": "ResembleAI/Dramabox",
44
+ "weight_format": "split_safetensors",
45
+ "watermarking": "skipped"
46
+ }
47
+ }
model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8ef8f665c966f526411a328a56c95bf71ec3c1adb8f19d9af09c24d42b7c5622
3
+ size 6575182591
model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2f8011adc777cacebceb3430e427954399563247508edcd3f67ddec8c0b34508
3
+ size 1942799557