rampasek commited on
Commit
5e48e4f
1 Parent(s): bf4a079
README.md ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # ProtBert-BFD finetuned on Rosetta 20,40,60AA dataset
2
+
3
+ This model is finetuned to predict Rosetta fold energy using a dataset of 300k protein sequences:
4
+ 100k of 20AA, 100k of 40AA, and 100k of 60AA
5
+
6
+ Current model in this repo: `prot_bert_bfd-finetuned-032822_1323`
7
+
8
+ ## Performance
9
+
10
+ On a held-out eval set the performance is:
11
+ - 20AA sequences (1k eval set):
12
+ Metrics: 'mae': 0.100418, 'r2': 0.989028, 'mse': 0.016266, 'rmse': 0.127537
13
+ - 40AA sequences (10k eval set):
14
+ Metrics: 'mae': 0.173888, 'r2': 0.963361, 'mse': 0.048218, 'rmse': 0.219587
15
+ - 60AA sequences (10k eval set):
16
+ Metrics: 'mae': 0.235238, 'r2': 0.930164, 'mse': 0.088131, 'rmse': 0.2968
17
+
18
+ ## `prot_bert_bfd` from ProtTrans
19
+ The starting pretrained model is from ProtTrans, trained on 2.1 billion proteins from BFD.
20
+ It was trained on protein sequences using a masked language modeling (MLM) objective. It was introduced in
21
+ [this paper](https://doi.org/10.1101/2020.07.12.199554) and first released in
22
+ [this repository](https://github.com/agemagician/ProtTrans).
23
+
config.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "Rostlab/prot_bert_bfd",
3
+ "architectures": [
4
+ "BertForSequenceClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.0,
7
+ "classifier_dropout": null,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 1024,
11
+ "id2label": {
12
+ "0": "LABEL_0"
13
+ },
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 4096,
16
+ "label2id": {
17
+ "LABEL_0": 0
18
+ },
19
+ "layer_norm_eps": 1e-12,
20
+ "max_position_embeddings": 40000,
21
+ "model_type": "bert",
22
+ "num_attention_heads": 16,
23
+ "num_hidden_layers": 30,
24
+ "pad_token_id": 0,
25
+ "position_embedding_type": "absolute",
26
+ "problem_type": "regression",
27
+ "torch_dtype": "float32",
28
+ "transformers_version": "4.17.0",
29
+ "type_vocab_size": 2,
30
+ "use_cache": true,
31
+ "vocab_size": 30
32
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12d7ef6b70b6d2aec36a03d432f4f29a324823538b2d3761e0893273b87ae96d
3
+ size 1680247213
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}
tokenizer.json ADDED
@@ -0,0 +1,180 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
5
+ "added_tokens": [
6
+ {
7
+ "id": 0,
8
+ "content": "[PAD]",
9
+ "single_word": false,
10
+ "lstrip": false,
11
+ "rstrip": false,
12
+ "normalized": false,
13
+ "special": true
14
+ },
15
+ {
16
+ "id": 1,
17
+ "content": "[UNK]",
18
+ "single_word": false,
19
+ "lstrip": false,
20
+ "rstrip": false,
21
+ "normalized": false,
22
+ "special": true
23
+ },
24
+ {
25
+ "id": 2,
26
+ "content": "[CLS]",
27
+ "single_word": false,
28
+ "lstrip": false,
29
+ "rstrip": false,
30
+ "normalized": false,
31
+ "special": true
32
+ },
33
+ {
34
+ "id": 3,
35
+ "content": "[SEP]",
36
+ "single_word": false,
37
+ "lstrip": false,
38
+ "rstrip": false,
39
+ "normalized": false,
40
+ "special": true
41
+ },
42
+ {
43
+ "id": 4,
44
+ "content": "[MASK]",
45
+ "single_word": false,
46
+ "lstrip": false,
47
+ "rstrip": false,
48
+ "normalized": false,
49
+ "special": true
50
+ }
51
+ ],
52
+ "normalizer": {
53
+ "type": "BertNormalizer",
54
+ "clean_text": true,
55
+ "handle_chinese_chars": true,
56
+ "strip_accents": null,
57
+ "lowercase": false
58
+ },
59
+ "pre_tokenizer": {
60
+ "type": "BertPreTokenizer"
61
+ },
62
+ "post_processor": {
63
+ "type": "TemplateProcessing",
64
+ "single": [
65
+ {
66
+ "SpecialToken": {
67
+ "id": "[CLS]",
68
+ "type_id": 0
69
+ }
70
+ },
71
+ {
72
+ "Sequence": {
73
+ "id": "A",
74
+ "type_id": 0
75
+ }
76
+ },
77
+ {
78
+ "SpecialToken": {
79
+ "id": "[SEP]",
80
+ "type_id": 0
81
+ }
82
+ }
83
+ ],
84
+ "pair": [
85
+ {
86
+ "SpecialToken": {
87
+ "id": "[CLS]",
88
+ "type_id": 0
89
+ }
90
+ },
91
+ {
92
+ "Sequence": {
93
+ "id": "A",
94
+ "type_id": 0
95
+ }
96
+ },
97
+ {
98
+ "SpecialToken": {
99
+ "id": "[SEP]",
100
+ "type_id": 0
101
+ }
102
+ },
103
+ {
104
+ "Sequence": {
105
+ "id": "B",
106
+ "type_id": 1
107
+ }
108
+ },
109
+ {
110
+ "SpecialToken": {
111
+ "id": "[SEP]",
112
+ "type_id": 1
113
+ }
114
+ }
115
+ ],
116
+ "special_tokens": {
117
+ "[CLS]": {
118
+ "id": "[CLS]",
119
+ "ids": [
120
+ 2
121
+ ],
122
+ "tokens": [
123
+ "[CLS]"
124
+ ]
125
+ },
126
+ "[SEP]": {
127
+ "id": "[SEP]",
128
+ "ids": [
129
+ 3
130
+ ],
131
+ "tokens": [
132
+ "[SEP]"
133
+ ]
134
+ }
135
+ }
136
+ },
137
+ "decoder": {
138
+ "type": "WordPiece",
139
+ "prefix": "##",
140
+ "cleanup": true
141
+ },
142
+ "model": {
143
+ "type": "WordPiece",
144
+ "unk_token": "[UNK]",
145
+ "continuing_subword_prefix": "##",
146
+ "max_input_chars_per_word": 100,
147
+ "vocab": {
148
+ "[PAD]": 0,
149
+ "[UNK]": 1,
150
+ "[CLS]": 2,
151
+ "[SEP]": 3,
152
+ "[MASK]": 4,
153
+ "L": 5,
154
+ "A": 6,
155
+ "G": 7,
156
+ "V": 8,
157
+ "E": 9,
158
+ "S": 10,
159
+ "I": 11,
160
+ "K": 12,
161
+ "R": 13,
162
+ "D": 14,
163
+ "T": 15,
164
+ "P": 16,
165
+ "N": 17,
166
+ "Q": 18,
167
+ "F": 19,
168
+ "Y": 20,
169
+ "M": 21,
170
+ "H": 22,
171
+ "C": 23,
172
+ "W": 24,
173
+ "X": 25,
174
+ "U": 26,
175
+ "B": 27,
176
+ "Z": 28,
177
+ "O": 29
178
+ }
179
+ }
180
+ }
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"do_lower_case": false, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "special_tokens_map_file": null, "full_tokenizer_file": null, "name_or_path": "Rostlab/prot_bert_bfd", "do_basic_tokenize": true, "never_split": null, "tokenizer_class": "BertTokenizer"}
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2e28fb0e1bc54eb95ba166aaa5181e97f33801131164088bd55fea45251f95eb
3
+ size 2991
vocab.txt ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [PAD]
2
+ [UNK]
3
+ [CLS]
4
+ [SEP]
5
+ [MASK]
6
+ L
7
+ A
8
+ G
9
+ V
10
+ E
11
+ S
12
+ I
13
+ K
14
+ R
15
+ D
16
+ T
17
+ P
18
+ N
19
+ Q
20
+ F
21
+ Y
22
+ M
23
+ H
24
+ C
25
+ W
26
+ X
27
+ U
28
+ B
29
+ Z
30
+ O