Upload folder using huggingface_hub
Browse files- config.json +41 -0
- model.safetensors +3 -0
config.json
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"S3GenModel"
|
| 4 |
+
],
|
| 5 |
+
"cfm_inference_cfg_rate": 0.7,
|
| 6 |
+
"cfm_sigma_min": 1e-06,
|
| 7 |
+
"cfm_solver": "euler",
|
| 8 |
+
"cfm_t_scheduler": "cosine",
|
| 9 |
+
"decoder_act_fn": "gelu",
|
| 10 |
+
"decoder_attention_head_dim": 64,
|
| 11 |
+
"decoder_channels": [
|
| 12 |
+
256
|
| 13 |
+
],
|
| 14 |
+
"decoder_in_channels": 320,
|
| 15 |
+
"decoder_n_blocks": 4,
|
| 16 |
+
"decoder_num_heads": 8,
|
| 17 |
+
"decoder_num_mid_blocks": 12,
|
| 18 |
+
"decoder_out_channels": 80,
|
| 19 |
+
"dtype": "float32",
|
| 20 |
+
"encoder_attention_heads": 8,
|
| 21 |
+
"encoder_dropout_rate": 0.1,
|
| 22 |
+
"encoder_linear_units": 2048,
|
| 23 |
+
"encoder_num_blocks": 6,
|
| 24 |
+
"encoder_output_size": 512,
|
| 25 |
+
"fmax": 8000,
|
| 26 |
+
"fmin": 0,
|
| 27 |
+
"hop_length": 480,
|
| 28 |
+
"input_frame_rate": 25,
|
| 29 |
+
"mel_bins": 80,
|
| 30 |
+
"model_type": "s3gen",
|
| 31 |
+
"n_fft": 1920,
|
| 32 |
+
"pre_lookahead_len": 3,
|
| 33 |
+
"sampling_rate": 24000,
|
| 34 |
+
"speaker_embed_dim": 192,
|
| 35 |
+
"speaker_feat_dim": 80,
|
| 36 |
+
"token_embed_dim": 512,
|
| 37 |
+
"token_mel_ratio": 2,
|
| 38 |
+
"transformers_version": "5.0.0.dev0",
|
| 39 |
+
"vocab_size": 6561,
|
| 40 |
+
"win_size": 1920
|
| 41 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:457cee5eb88811beff5993cf01d3b16e731ae7969104f89b89438ae10c88b6b7
|
| 3 |
+
size 1056487412
|