Upload model
Browse files- config.json +16 -1
- generation_config.json +7 -0
- model-00001-of-00002.safetensors +1 -1
- model-00002-of-00002.safetensors +1 -1
config.json
CHANGED
|
@@ -2,6 +2,7 @@
|
|
| 2 |
"architectures": [
|
| 3 |
"DiaModel"
|
| 4 |
],
|
|
|
|
| 5 |
"decoder_config": {
|
| 6 |
"_attn_implementation_autoset": true,
|
| 7 |
"cross_head_dim": 128,
|
|
@@ -20,8 +21,20 @@
|
|
| 20 |
"num_key_value_heads": 4,
|
| 21 |
"rope_max_timescale": 10000,
|
| 22 |
"rope_min_timescale": 1,
|
|
|
|
| 23 |
"vocab_size": 1028
|
| 24 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
"encoder_config": {
|
| 26 |
"_attn_implementation_autoset": true,
|
| 27 |
"dropout": 0,
|
|
@@ -36,12 +49,14 @@
|
|
| 36 |
"num_key_value_heads": 16,
|
| 37 |
"rope_max_timescale": 10000,
|
| 38 |
"rope_min_timescale": 1,
|
|
|
|
| 39 |
"vocab_size": 256
|
| 40 |
},
|
|
|
|
| 41 |
"is_encoder_decoder": true,
|
| 42 |
"model_type": "dia",
|
| 43 |
"norm_eps": 1e-05,
|
| 44 |
-
"pad_token_id":
|
| 45 |
"torch_dtype": "float32",
|
| 46 |
"transformers_version": "4.52.0.dev0"
|
| 47 |
}
|
|
|
|
| 2 |
"architectures": [
|
| 3 |
"DiaModel"
|
| 4 |
],
|
| 5 |
+
"bos_token_id": 1026,
|
| 6 |
"decoder_config": {
|
| 7 |
"_attn_implementation_autoset": true,
|
| 8 |
"cross_head_dim": 128,
|
|
|
|
| 21 |
"num_key_value_heads": 4,
|
| 22 |
"rope_max_timescale": 10000,
|
| 23 |
"rope_min_timescale": 1,
|
| 24 |
+
"torch_dtype": "float32",
|
| 25 |
"vocab_size": 1028
|
| 26 |
},
|
| 27 |
+
"delay_pattern": [
|
| 28 |
+
0,
|
| 29 |
+
8,
|
| 30 |
+
9,
|
| 31 |
+
10,
|
| 32 |
+
11,
|
| 33 |
+
12,
|
| 34 |
+
13,
|
| 35 |
+
14,
|
| 36 |
+
15
|
| 37 |
+
],
|
| 38 |
"encoder_config": {
|
| 39 |
"_attn_implementation_autoset": true,
|
| 40 |
"dropout": 0,
|
|
|
|
| 49 |
"num_key_value_heads": 16,
|
| 50 |
"rope_max_timescale": 10000,
|
| 51 |
"rope_min_timescale": 1,
|
| 52 |
+
"torch_dtype": "float32",
|
| 53 |
"vocab_size": 256
|
| 54 |
},
|
| 55 |
+
"eos_token_id": 1024,
|
| 56 |
"is_encoder_decoder": true,
|
| 57 |
"model_type": "dia",
|
| 58 |
"norm_eps": 1e-05,
|
| 59 |
+
"pad_token_id": 1025,
|
| 60 |
"torch_dtype": "float32",
|
| 61 |
"transformers_version": "4.52.0.dev0"
|
| 62 |
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1026,
|
| 4 |
+
"eos_token_id": 1024,
|
| 5 |
+
"pad_token_id": 1025,
|
| 6 |
+
"transformers_version": "4.52.0.dev0"
|
| 7 |
+
}
|
model-00001-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4993044448
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2c7d787ff2bc1f85dcdb74e4ccfdc9dd7a59c03bcd5df0f3b534f17894eaed77
|
| 3 |
size 4993044448
|
model-00002-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1451637168
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b9b3eed74c98c55f0b4f3ba3f8c3ccb65716f9f3c48bf1bda72f1433b95f27a
|
| 3 |
size 1451637168
|