update config.json
Browse files- config.json +53 -5
config.json
CHANGED
|
@@ -1,6 +1,54 @@
|
|
| 1 |
-
{
|
| 2 |
-
"model_type": "dia",
|
| 3 |
-
"quantized": true,
|
| 4 |
-
"quantization_type": "dynamic",
|
| 5 |
-
"original_model": "nari-labs/Dia-1.6B"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
}
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_type": "dia",
|
| 3 |
+
"quantized": true,
|
| 4 |
+
"quantization_type": "dynamic",
|
| 5 |
+
"original_model": "nari-labs/Dia-1.6B",
|
| 6 |
+
"data": {
|
| 7 |
+
"audio_bos_value": 1026,
|
| 8 |
+
"audio_eos_value": 1024,
|
| 9 |
+
"audio_length": 3072,
|
| 10 |
+
"audio_pad_value": 1025,
|
| 11 |
+
"channels": 9,
|
| 12 |
+
"delay_pattern": [
|
| 13 |
+
0,
|
| 14 |
+
8,
|
| 15 |
+
9,
|
| 16 |
+
10,
|
| 17 |
+
11,
|
| 18 |
+
12,
|
| 19 |
+
13,
|
| 20 |
+
14,
|
| 21 |
+
15
|
| 22 |
+
],
|
| 23 |
+
"text_length": 1024,
|
| 24 |
+
"text_pad_value": 0
|
| 25 |
+
},
|
| 26 |
+
"model": {
|
| 27 |
+
"decoder": {
|
| 28 |
+
"cross_head_dim": 128,
|
| 29 |
+
"cross_query_heads": 16,
|
| 30 |
+
"gqa_head_dim": 128,
|
| 31 |
+
"gqa_query_heads": 16,
|
| 32 |
+
"kv_heads": 4,
|
| 33 |
+
"n_embd": 2048,
|
| 34 |
+
"n_hidden": 8192,
|
| 35 |
+
"n_layer": 18
|
| 36 |
+
},
|
| 37 |
+
"dropout": 0.0,
|
| 38 |
+
"encoder": {
|
| 39 |
+
"head_dim": 128,
|
| 40 |
+
"n_embd": 1024,
|
| 41 |
+
"n_head": 16,
|
| 42 |
+
"n_hidden": 4096,
|
| 43 |
+
"n_layer": 12
|
| 44 |
+
},
|
| 45 |
+
"normalization_layer_epsilon": 1e-05,
|
| 46 |
+
"rope_max_timescale": 10000,
|
| 47 |
+
"rope_min_timescale": 1,
|
| 48 |
+
"src_vocab_size": 256,
|
| 49 |
+
"tgt_vocab_size": 1028,
|
| 50 |
+
"weight_dtype": "float32"
|
| 51 |
+
},
|
| 52 |
+
"training": {},
|
| 53 |
+
"version": "0.1"
|
| 54 |
}
|