patrickvonplaten
commited on
Commit
•
82478a1
1
Parent(s):
7613e44
upload
Browse files
bert/config.json
ADDED
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "hf-internal-testing/tiny-random-bert",
|
3 |
+
"architectures": [
|
4 |
+
"BertModel"
|
5 |
+
],
|
6 |
+
"attention_probs_dropout_prob": 0.1,
|
7 |
+
"classifier_dropout": null,
|
8 |
+
"gradient_checkpointing": false,
|
9 |
+
"hidden_act": "gelu",
|
10 |
+
"hidden_dropout_prob": 0.1,
|
11 |
+
"hidden_size": 32,
|
12 |
+
"initializer_range": 0.02,
|
13 |
+
"intermediate_size": 37,
|
14 |
+
"layer_norm_eps": 1e-12,
|
15 |
+
"max_position_embeddings": 512,
|
16 |
+
"model_type": "bert",
|
17 |
+
"num_attention_heads": 4,
|
18 |
+
"num_hidden_layers": 5,
|
19 |
+
"pad_token_id": 0,
|
20 |
+
"position_embedding_type": "absolute",
|
21 |
+
"torch_dtype": "float32",
|
22 |
+
"transformers_version": "4.18.0.dev0",
|
23 |
+
"type_vocab_size": 16,
|
24 |
+
"use_cache": true,
|
25 |
+
"vocab_size": 1124
|
26 |
+
}
|
bert/pytorch_model-00001-of-00005.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bd420eb358a99a5150cb374dec757373d5cc91b0b460ac2eb2d3bfb22888dcf6
|
3 |
+
size 4843
|
bert/pytorch_model-00002-of-00005.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a37b91bed13a298130eb8a8a8ae19477cc5ddb6ebd0ee89abe109cb3474d23a7
|
3 |
+
size 144683
|
bert/pytorch_model-00003-of-00005.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3c1ab90263be374c6c66cc3f368b943585fc8f40e59571c6c98e1664edea9690
|
3 |
+
size 105443
|
bert/pytorch_model-00004-of-00005.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f069b4362265840846cc44f3ad125893917fcaa23afa41bf5f7de0ab680969cc
|
3 |
+
size 116599
|
bert/pytorch_model-00005-of-00005.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a7fb234990f8f085471fe63b90c1039a48dc904762962d21928764fb91ee9234
|
3 |
+
size 11159
|
bert/pytorch_model.bin.index.json
ADDED
@@ -0,0 +1,96 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"dtype": "int64",
|
3 |
+
"metadata": {
|
4 |
+
"total_size": 355812
|
5 |
+
},
|
6 |
+
"weight_map": {
|
7 |
+
"embeddings.LayerNorm.bias": "pytorch_model-00003-of-00005.bin",
|
8 |
+
"embeddings.LayerNorm.weight": "pytorch_model-00003-of-00005.bin",
|
9 |
+
"embeddings.position_embeddings.weight": "pytorch_model-00003-of-00005.bin",
|
10 |
+
"embeddings.position_ids": "pytorch_model-00001-of-00005.bin",
|
11 |
+
"embeddings.token_type_embeddings.weight": "pytorch_model-00003-of-00005.bin",
|
12 |
+
"embeddings.word_embeddings.weight": "pytorch_model-00002-of-00005.bin",
|
13 |
+
"encoder.layer.0.attention.output.LayerNorm.bias": "pytorch_model-00003-of-00005.bin",
|
14 |
+
"encoder.layer.0.attention.output.LayerNorm.weight": "pytorch_model-00003-of-00005.bin",
|
15 |
+
"encoder.layer.0.attention.output.dense.bias": "pytorch_model-00003-of-00005.bin",
|
16 |
+
"encoder.layer.0.attention.output.dense.weight": "pytorch_model-00003-of-00005.bin",
|
17 |
+
"encoder.layer.0.attention.self.key.bias": "pytorch_model-00003-of-00005.bin",
|
18 |
+
"encoder.layer.0.attention.self.key.weight": "pytorch_model-00003-of-00005.bin",
|
19 |
+
"encoder.layer.0.attention.self.query.bias": "pytorch_model-00003-of-00005.bin",
|
20 |
+
"encoder.layer.0.attention.self.query.weight": "pytorch_model-00003-of-00005.bin",
|
21 |
+
"encoder.layer.0.attention.self.value.bias": "pytorch_model-00003-of-00005.bin",
|
22 |
+
"encoder.layer.0.attention.self.value.weight": "pytorch_model-00003-of-00005.bin",
|
23 |
+
"encoder.layer.0.intermediate.dense.bias": "pytorch_model-00003-of-00005.bin",
|
24 |
+
"encoder.layer.0.intermediate.dense.weight": "pytorch_model-00003-of-00005.bin",
|
25 |
+
"encoder.layer.0.output.LayerNorm.bias": "pytorch_model-00003-of-00005.bin",
|
26 |
+
"encoder.layer.0.output.LayerNorm.weight": "pytorch_model-00003-of-00005.bin",
|
27 |
+
"encoder.layer.0.output.dense.bias": "pytorch_model-00003-of-00005.bin",
|
28 |
+
"encoder.layer.0.output.dense.weight": "pytorch_model-00003-of-00005.bin",
|
29 |
+
"encoder.layer.1.attention.output.LayerNorm.bias": "pytorch_model-00004-of-00005.bin",
|
30 |
+
"encoder.layer.1.attention.output.LayerNorm.weight": "pytorch_model-00004-of-00005.bin",
|
31 |
+
"encoder.layer.1.attention.output.dense.bias": "pytorch_model-00004-of-00005.bin",
|
32 |
+
"encoder.layer.1.attention.output.dense.weight": "pytorch_model-00004-of-00005.bin",
|
33 |
+
"encoder.layer.1.attention.self.key.bias": "pytorch_model-00004-of-00005.bin",
|
34 |
+
"encoder.layer.1.attention.self.key.weight": "pytorch_model-00004-of-00005.bin",
|
35 |
+
"encoder.layer.1.attention.self.query.bias": "pytorch_model-00003-of-00005.bin",
|
36 |
+
"encoder.layer.1.attention.self.query.weight": "pytorch_model-00003-of-00005.bin",
|
37 |
+
"encoder.layer.1.attention.self.value.bias": "pytorch_model-00004-of-00005.bin",
|
38 |
+
"encoder.layer.1.attention.self.value.weight": "pytorch_model-00004-of-00005.bin",
|
39 |
+
"encoder.layer.1.intermediate.dense.bias": "pytorch_model-00004-of-00005.bin",
|
40 |
+
"encoder.layer.1.intermediate.dense.weight": "pytorch_model-00004-of-00005.bin",
|
41 |
+
"encoder.layer.1.output.LayerNorm.bias": "pytorch_model-00004-of-00005.bin",
|
42 |
+
"encoder.layer.1.output.LayerNorm.weight": "pytorch_model-00004-of-00005.bin",
|
43 |
+
"encoder.layer.1.output.dense.bias": "pytorch_model-00004-of-00005.bin",
|
44 |
+
"encoder.layer.1.output.dense.weight": "pytorch_model-00004-of-00005.bin",
|
45 |
+
"encoder.layer.2.attention.output.LayerNorm.bias": "pytorch_model-00004-of-00005.bin",
|
46 |
+
"encoder.layer.2.attention.output.LayerNorm.weight": "pytorch_model-00004-of-00005.bin",
|
47 |
+
"encoder.layer.2.attention.output.dense.bias": "pytorch_model-00004-of-00005.bin",
|
48 |
+
"encoder.layer.2.attention.output.dense.weight": "pytorch_model-00004-of-00005.bin",
|
49 |
+
"encoder.layer.2.attention.self.key.bias": "pytorch_model-00004-of-00005.bin",
|
50 |
+
"encoder.layer.2.attention.self.key.weight": "pytorch_model-00004-of-00005.bin",
|
51 |
+
"encoder.layer.2.attention.self.query.bias": "pytorch_model-00004-of-00005.bin",
|
52 |
+
"encoder.layer.2.attention.self.query.weight": "pytorch_model-00004-of-00005.bin",
|
53 |
+
"encoder.layer.2.attention.self.value.bias": "pytorch_model-00004-of-00005.bin",
|
54 |
+
"encoder.layer.2.attention.self.value.weight": "pytorch_model-00004-of-00005.bin",
|
55 |
+
"encoder.layer.2.intermediate.dense.bias": "pytorch_model-00004-of-00005.bin",
|
56 |
+
"encoder.layer.2.intermediate.dense.weight": "pytorch_model-00004-of-00005.bin",
|
57 |
+
"encoder.layer.2.output.LayerNorm.bias": "pytorch_model-00004-of-00005.bin",
|
58 |
+
"encoder.layer.2.output.LayerNorm.weight": "pytorch_model-00004-of-00005.bin",
|
59 |
+
"encoder.layer.2.output.dense.bias": "pytorch_model-00004-of-00005.bin",
|
60 |
+
"encoder.layer.2.output.dense.weight": "pytorch_model-00004-of-00005.bin",
|
61 |
+
"encoder.layer.3.attention.output.LayerNorm.bias": "pytorch_model-00004-of-00005.bin",
|
62 |
+
"encoder.layer.3.attention.output.LayerNorm.weight": "pytorch_model-00004-of-00005.bin",
|
63 |
+
"encoder.layer.3.attention.output.dense.bias": "pytorch_model-00004-of-00005.bin",
|
64 |
+
"encoder.layer.3.attention.output.dense.weight": "pytorch_model-00004-of-00005.bin",
|
65 |
+
"encoder.layer.3.attention.self.key.bias": "pytorch_model-00004-of-00005.bin",
|
66 |
+
"encoder.layer.3.attention.self.key.weight": "pytorch_model-00004-of-00005.bin",
|
67 |
+
"encoder.layer.3.attention.self.query.bias": "pytorch_model-00004-of-00005.bin",
|
68 |
+
"encoder.layer.3.attention.self.query.weight": "pytorch_model-00004-of-00005.bin",
|
69 |
+
"encoder.layer.3.attention.self.value.bias": "pytorch_model-00004-of-00005.bin",
|
70 |
+
"encoder.layer.3.attention.self.value.weight": "pytorch_model-00004-of-00005.bin",
|
71 |
+
"encoder.layer.3.intermediate.dense.bias": "pytorch_model-00004-of-00005.bin",
|
72 |
+
"encoder.layer.3.intermediate.dense.weight": "pytorch_model-00004-of-00005.bin",
|
73 |
+
"encoder.layer.3.output.LayerNorm.bias": "pytorch_model-00004-of-00005.bin",
|
74 |
+
"encoder.layer.3.output.LayerNorm.weight": "pytorch_model-00004-of-00005.bin",
|
75 |
+
"encoder.layer.3.output.dense.bias": "pytorch_model-00004-of-00005.bin",
|
76 |
+
"encoder.layer.3.output.dense.weight": "pytorch_model-00004-of-00005.bin",
|
77 |
+
"encoder.layer.4.attention.output.LayerNorm.bias": "pytorch_model-00004-of-00005.bin",
|
78 |
+
"encoder.layer.4.attention.output.LayerNorm.weight": "pytorch_model-00004-of-00005.bin",
|
79 |
+
"encoder.layer.4.attention.output.dense.bias": "pytorch_model-00004-of-00005.bin",
|
80 |
+
"encoder.layer.4.attention.output.dense.weight": "pytorch_model-00004-of-00005.bin",
|
81 |
+
"encoder.layer.4.attention.self.key.bias": "pytorch_model-00004-of-00005.bin",
|
82 |
+
"encoder.layer.4.attention.self.key.weight": "pytorch_model-00004-of-00005.bin",
|
83 |
+
"encoder.layer.4.attention.self.query.bias": "pytorch_model-00004-of-00005.bin",
|
84 |
+
"encoder.layer.4.attention.self.query.weight": "pytorch_model-00004-of-00005.bin",
|
85 |
+
"encoder.layer.4.attention.self.value.bias": "pytorch_model-00004-of-00005.bin",
|
86 |
+
"encoder.layer.4.attention.self.value.weight": "pytorch_model-00004-of-00005.bin",
|
87 |
+
"encoder.layer.4.intermediate.dense.bias": "pytorch_model-00004-of-00005.bin",
|
88 |
+
"encoder.layer.4.intermediate.dense.weight": "pytorch_model-00004-of-00005.bin",
|
89 |
+
"encoder.layer.4.output.LayerNorm.bias": "pytorch_model-00005-of-00005.bin",
|
90 |
+
"encoder.layer.4.output.LayerNorm.weight": "pytorch_model-00005-of-00005.bin",
|
91 |
+
"encoder.layer.4.output.dense.bias": "pytorch_model-00005-of-00005.bin",
|
92 |
+
"encoder.layer.4.output.dense.weight": "pytorch_model-00005-of-00005.bin",
|
93 |
+
"pooler.dense.bias": "pytorch_model-00005-of-00005.bin",
|
94 |
+
"pooler.dense.weight": "pytorch_model-00005-of-00005.bin"
|
95 |
+
}
|
96 |
+
}
|