mokcho commited on
Commit
ca5c18a
·
verified ·
1 Parent(s): e49f82a

Upload folder using huggingface_hub

Browse files
East_Slavic/UD_Russian-SynTagRus/config.json ADDED
@@ -0,0 +1,187 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "bert-base-multilingual-cased",
3
+ "adapters": {
4
+ "adapters": {},
5
+ "config_map": {},
6
+ "fusion_config_map": {},
7
+ "fusion_name_map": {},
8
+ "fusions": {}
9
+ },
10
+ "architectures": [
11
+ "BertAdapterModel"
12
+ ],
13
+ "attention_probs_dropout_prob": 0.1,
14
+ "classifier_dropout": null,
15
+ "directionality": "bidi",
16
+ "hidden_act": "gelu",
17
+ "hidden_dropout_prob": 0.1,
18
+ "hidden_size": 768,
19
+ "id2label": {
20
+ "0": "_",
21
+ "1": "acl",
22
+ "2": "advcl",
23
+ "3": "advmod",
24
+ "4": "amod",
25
+ "5": "appos",
26
+ "6": "aux",
27
+ "7": "case",
28
+ "8": "cc",
29
+ "9": "ccomp",
30
+ "10": "clf",
31
+ "11": "compound",
32
+ "12": "conj",
33
+ "13": "cop",
34
+ "14": "csubj",
35
+ "15": "dep",
36
+ "16": "det",
37
+ "17": "discourse",
38
+ "18": "dislocated",
39
+ "19": "expl",
40
+ "20": "fixed",
41
+ "21": "flat",
42
+ "22": "goeswith",
43
+ "23": "iobj",
44
+ "24": "list",
45
+ "25": "mark",
46
+ "26": "nmod",
47
+ "27": "nsubj",
48
+ "28": "nummod",
49
+ "29": "obj",
50
+ "30": "obl",
51
+ "31": "orphan",
52
+ "32": "parataxis",
53
+ "33": "punct",
54
+ "34": "reparandum",
55
+ "35": "root",
56
+ "36": "vocative",
57
+ "37": "xcomp",
58
+ "38": "dup",
59
+ "39": "mwe",
60
+ "40": "name",
61
+ "41": "remnant"
62
+ },
63
+ "initializer_range": 0.02,
64
+ "intermediate_size": 3072,
65
+ "label2id": {
66
+ "_": 0,
67
+ "acl": 1,
68
+ "advcl": 2,
69
+ "advmod": 3,
70
+ "amod": 4,
71
+ "appos": 5,
72
+ "aux": 6,
73
+ "case": 7,
74
+ "cc": 8,
75
+ "ccomp": 9,
76
+ "clf": 10,
77
+ "compound": 11,
78
+ "conj": 12,
79
+ "cop": 13,
80
+ "csubj": 14,
81
+ "dep": 15,
82
+ "det": 16,
83
+ "discourse": 17,
84
+ "dislocated": 18,
85
+ "dup": 38,
86
+ "expl": 19,
87
+ "fixed": 20,
88
+ "flat": 21,
89
+ "goeswith": 22,
90
+ "iobj": 23,
91
+ "list": 24,
92
+ "mark": 25,
93
+ "mwe": 39,
94
+ "name": 40,
95
+ "nmod": 26,
96
+ "nsubj": 27,
97
+ "nummod": 28,
98
+ "obj": 29,
99
+ "obl": 30,
100
+ "orphan": 31,
101
+ "parataxis": 32,
102
+ "punct": 33,
103
+ "remnant": 41,
104
+ "reparandum": 34,
105
+ "root": 35,
106
+ "vocative": 36,
107
+ "xcomp": 37
108
+ },
109
+ "layer_norm_eps": 1e-12,
110
+ "max_position_embeddings": 512,
111
+ "model_type": "bert",
112
+ "num_attention_heads": 12,
113
+ "num_hidden_layers": 12,
114
+ "pad_token_id": 0,
115
+ "pooler_fc_size": 768,
116
+ "pooler_num_attention_heads": 12,
117
+ "pooler_num_fc_layers": 3,
118
+ "pooler_size_per_head": 128,
119
+ "pooler_type": "first_token_transform",
120
+ "position_embedding_type": "absolute",
121
+ "prediction_heads": {
122
+ "default": {
123
+ "activation_function": "gelu",
124
+ "bias": true,
125
+ "embedding_size": 768,
126
+ "head_type": "masked_lm",
127
+ "label2id": null,
128
+ "layer_norm": true,
129
+ "layers": 2,
130
+ "shift_labels": false,
131
+ "vocab_size": 119547
132
+ },
133
+ "ud_UD_Russian-SynTagRus": {
134
+ "head_type": "dependency_parsing",
135
+ "label2id": {
136
+ "_": 0,
137
+ "acl": 1,
138
+ "advcl": 2,
139
+ "advmod": 3,
140
+ "amod": 4,
141
+ "appos": 5,
142
+ "aux": 6,
143
+ "case": 7,
144
+ "cc": 8,
145
+ "ccomp": 9,
146
+ "clf": 10,
147
+ "compound": 11,
148
+ "conj": 12,
149
+ "cop": 13,
150
+ "csubj": 14,
151
+ "dep": 15,
152
+ "det": 16,
153
+ "discourse": 17,
154
+ "dislocated": 18,
155
+ "dup": 38,
156
+ "expl": 19,
157
+ "fixed": 20,
158
+ "flat": 21,
159
+ "goeswith": 22,
160
+ "iobj": 23,
161
+ "list": 24,
162
+ "mark": 25,
163
+ "mwe": 39,
164
+ "name": 40,
165
+ "nmod": 26,
166
+ "nsubj": 27,
167
+ "nummod": 28,
168
+ "obj": 29,
169
+ "obl": 30,
170
+ "orphan": 31,
171
+ "parataxis": 32,
172
+ "punct": 33,
173
+ "remnant": 41,
174
+ "reparandum": 34,
175
+ "root": 35,
176
+ "vocative": 36,
177
+ "xcomp": 37
178
+ },
179
+ "num_labels": 42
180
+ }
181
+ },
182
+ "torch_dtype": "float32",
183
+ "transformers_version": "4.47.1",
184
+ "type_vocab_size": 2,
185
+ "use_cache": true,
186
+ "vocab_size": 119547
187
+ }
East_Slavic/UD_Russian-SynTagRus/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b8b07d541ace3f22f7e7aedd7d44dbfd8691c73053f82e6d2e2ba69323f06118
3
+ size 815995564
East_Slavic/UD_Russian-SynTagRus/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:402cd545f73a42507cb5e7cbc6f284c10564b148e39f5fb72de1c966b5a8cf51
3
+ size 1632115130
East_Slavic/UD_Russian-SynTagRus/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af4a7b8bad5d6db8e6d9392f9a72ed51b01d8583a5e69d10403e5e82aaff3cb3
3
+ size 14244
East_Slavic/UD_Russian-SynTagRus/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:434b0e57ebc0438ac4e4adbd6f8f912ad68b0609249fd8062ce755a4e58e1082
3
+ size 1064
East_Slavic/UD_Russian-SynTagRus/trainer_state.json ADDED
@@ -0,0 +1,131 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 5.0,
5
+ "eval_steps": 500,
6
+ "global_step": 5440,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.45955882352941174,
13
+ "grad_norm": 3.0535507202148438,
14
+ "learning_rate": 0.00018161764705882353,
15
+ "loss": 0.7526,
16
+ "step": 500
17
+ },
18
+ {
19
+ "epoch": 0.9191176470588235,
20
+ "grad_norm": 2.3940510749816895,
21
+ "learning_rate": 0.00016323529411764706,
22
+ "loss": 0.4683,
23
+ "step": 1000
24
+ },
25
+ {
26
+ "epoch": 1.0,
27
+ "eval_loss": 0.4154830680661826,
28
+ "las": 88.15367234771037,
29
+ "step": 1088,
30
+ "uas": 91.78062253519175
31
+ },
32
+ {
33
+ "epoch": 1.3786764705882353,
34
+ "grad_norm": 1.989762306213379,
35
+ "learning_rate": 0.00014485294117647058,
36
+ "loss": 0.3564,
37
+ "step": 1500
38
+ },
39
+ {
40
+ "epoch": 1.8382352941176472,
41
+ "grad_norm": 1.942738652229309,
42
+ "learning_rate": 0.0001264705882352941,
43
+ "loss": 0.3265,
44
+ "step": 2000
45
+ },
46
+ {
47
+ "epoch": 2.0,
48
+ "eval_loss": 0.35586839087668615,
49
+ "las": 89.64867168927302,
50
+ "step": 2176,
51
+ "uas": 92.89164893506423
52
+ },
53
+ {
54
+ "epoch": 2.297794117647059,
55
+ "grad_norm": 1.5712599754333496,
56
+ "learning_rate": 0.00010808823529411764,
57
+ "loss": 0.2617,
58
+ "step": 2500
59
+ },
60
+ {
61
+ "epoch": 2.7573529411764706,
62
+ "grad_norm": 1.599707007408142,
63
+ "learning_rate": 8.970588235294118e-05,
64
+ "loss": 0.2296,
65
+ "step": 3000
66
+ },
67
+ {
68
+ "epoch": 3.0,
69
+ "eval_loss": 0.34188422671046037,
70
+ "las": 90.35493238888004,
71
+ "step": 3264,
72
+ "uas": 93.37057547424817
73
+ },
74
+ {
75
+ "epoch": 3.2169117647058822,
76
+ "grad_norm": 2.7295961380004883,
77
+ "learning_rate": 7.13235294117647e-05,
78
+ "loss": 0.1911,
79
+ "step": 3500
80
+ },
81
+ {
82
+ "epoch": 3.6764705882352944,
83
+ "grad_norm": 1.7654919624328613,
84
+ "learning_rate": 5.294117647058824e-05,
85
+ "loss": 0.1525,
86
+ "step": 4000
87
+ },
88
+ {
89
+ "epoch": 4.0,
90
+ "eval_loss": 0.3498880854093836,
91
+ "las": 90.84078984758908,
92
+ "step": 4352,
93
+ "uas": 93.73375565736306
94
+ },
95
+ {
96
+ "epoch": 4.136029411764706,
97
+ "grad_norm": 4.418405532836914,
98
+ "learning_rate": 3.455882352941177e-05,
99
+ "loss": 0.1289,
100
+ "step": 4500
101
+ },
102
+ {
103
+ "epoch": 4.595588235294118,
104
+ "grad_norm": 1.2796601057052612,
105
+ "learning_rate": 1.6176470588235296e-05,
106
+ "loss": 0.0888,
107
+ "step": 5000
108
+ }
109
+ ],
110
+ "logging_steps": 500,
111
+ "max_steps": 5440,
112
+ "num_input_tokens_seen": 0,
113
+ "num_train_epochs": 5,
114
+ "save_steps": 500,
115
+ "stateful_callbacks": {
116
+ "TrainerControl": {
117
+ "args": {
118
+ "should_epoch_stop": false,
119
+ "should_evaluate": false,
120
+ "should_log": false,
121
+ "should_save": true,
122
+ "should_training_stop": true
123
+ },
124
+ "attributes": {}
125
+ }
126
+ },
127
+ "total_flos": 5.977593818240256e+16,
128
+ "train_batch_size": 64,
129
+ "trial_name": null,
130
+ "trial_params": null
131
+ }
East_Slavic/UD_Russian-SynTagRus/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c36bf2eba9180b2dcbff3938266e9e99528227255dca9cc608e8acdd4d2807c5
3
+ size 5432