patrickvonplaten
commited on
Commit
•
5a16a0c
1
Parent(s):
f91ea52
up
Browse files- config.json +23 -32
- pytorch_model.bin +2 -2
config.json
CHANGED
@@ -11,47 +11,38 @@
|
|
11 |
"contrastive_logits_temperature": 0.1,
|
12 |
"conv_bias": false,
|
13 |
"conv_dim": [
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
512,
|
18 |
-
512,
|
19 |
-
512,
|
20 |
-
512
|
21 |
],
|
22 |
"conv_kernel": [
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
3,
|
27 |
-
3,
|
28 |
-
2,
|
29 |
-
2
|
30 |
],
|
31 |
"conv_stride": [
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
2,
|
36 |
-
2,
|
37 |
-
2,
|
38 |
-
2
|
39 |
],
|
40 |
"ctc_loss_reduction": "sum",
|
41 |
"ctc_zero_infinity": false,
|
42 |
"diversity_loss_weight": 0.1,
|
43 |
-
"do_stable_layer_norm":
|
44 |
"eos_token_id": 2,
|
45 |
"feat_extract_activation": "gelu",
|
46 |
-
"
|
47 |
-
"
|
|
|
48 |
"feat_quantizer_dropout": 0.0,
|
49 |
"final_dropout": 0.1,
|
|
|
50 |
"hidden_act": "gelu",
|
51 |
"hidden_dropout": 0.1,
|
52 |
-
"
|
|
|
53 |
"initializer_range": 0.02,
|
54 |
-
"intermediate_size":
|
55 |
"layer_norm_eps": 1e-05,
|
56 |
"layerdrop": 0.1,
|
57 |
"mask_feature_length": 10,
|
@@ -59,15 +50,15 @@
|
|
59 |
"mask_time_length": 10,
|
60 |
"mask_time_prob": 0.05,
|
61 |
"model_type": "unispeech-sat",
|
62 |
-
"num_attention_heads":
|
63 |
"num_clusters": 504,
|
64 |
"num_codevector_groups": 2,
|
65 |
"num_codevectors_per_group": 320,
|
66 |
-
"num_conv_pos_embedding_groups":
|
67 |
-
"num_conv_pos_embeddings":
|
68 |
-
"num_feat_extract_layers":
|
69 |
-
"num_hidden_layers":
|
70 |
-
"num_negatives":
|
71 |
"pad_token_id": 0,
|
72 |
"proj_codevector_dim": 256,
|
73 |
"replace_prob": 0.5,
|
|
|
11 |
"contrastive_logits_temperature": 0.1,
|
12 |
"conv_bias": false,
|
13 |
"conv_dim": [
|
14 |
+
32,
|
15 |
+
32,
|
16 |
+
32
|
|
|
|
|
|
|
|
|
17 |
],
|
18 |
"conv_kernel": [
|
19 |
+
8,
|
20 |
+
8,
|
21 |
+
8
|
|
|
|
|
|
|
|
|
22 |
],
|
23 |
"conv_stride": [
|
24 |
+
4,
|
25 |
+
4,
|
26 |
+
4
|
|
|
|
|
|
|
|
|
27 |
],
|
28 |
"ctc_loss_reduction": "sum",
|
29 |
"ctc_zero_infinity": false,
|
30 |
"diversity_loss_weight": 0.1,
|
31 |
+
"do_stable_layer_norm": true,
|
32 |
"eos_token_id": 2,
|
33 |
"feat_extract_activation": "gelu",
|
34 |
+
"feat_extract_dropout": 0.0,
|
35 |
+
"feat_extract_norm": "layer",
|
36 |
+
"feat_proj_dropout": 0.1,
|
37 |
"feat_quantizer_dropout": 0.0,
|
38 |
"final_dropout": 0.1,
|
39 |
+
"gradient_checkpointing": false,
|
40 |
"hidden_act": "gelu",
|
41 |
"hidden_dropout": 0.1,
|
42 |
+
"hidden_dropout_prob": 0.1,
|
43 |
+
"hidden_size": 16,
|
44 |
"initializer_range": 0.02,
|
45 |
+
"intermediate_size": 20,
|
46 |
"layer_norm_eps": 1e-05,
|
47 |
"layerdrop": 0.1,
|
48 |
"mask_feature_length": 10,
|
|
|
50 |
"mask_time_length": 10,
|
51 |
"mask_time_prob": 0.05,
|
52 |
"model_type": "unispeech-sat",
|
53 |
+
"num_attention_heads": 2,
|
54 |
"num_clusters": 504,
|
55 |
"num_codevector_groups": 2,
|
56 |
"num_codevectors_per_group": 320,
|
57 |
+
"num_conv_pos_embedding_groups": 2,
|
58 |
+
"num_conv_pos_embeddings": 16,
|
59 |
+
"num_feat_extract_layers": 3,
|
60 |
+
"num_hidden_layers": 4,
|
61 |
+
"num_negatives": 10,
|
62 |
"pad_token_id": 0,
|
63 |
"proj_codevector_dim": 256,
|
64 |
"replace_prob": 0.5,
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b23b6f9effbd1924fd270668aa133e69f414d4bb5b4647f2096a37cdfe2ddb49
|
3 |
+
size 144684
|