asahi417 commited on
Commit
0e70593
1 Parent(s): 3da1da0
README.md CHANGED
@@ -14,7 +14,7 @@ model-index:
14
  metrics:
15
  - name: Accuracy
16
  type: accuracy
17
- value: 0.8133333333333334
18
  - task:
19
  name: Analogy Questions (SAT full)
20
  type: multiple-choice-qa
@@ -25,7 +25,7 @@ model-index:
25
  metrics:
26
  - name: Accuracy
27
  type: accuracy
28
- value: 0.6818181818181818
29
  - task:
30
  name: Analogy Questions (SAT)
31
  type: multiple-choice-qa
@@ -36,7 +36,7 @@ model-index:
36
  metrics:
37
  - name: Accuracy
38
  type: accuracy
39
- value: 0.6824925816023739
40
  - task:
41
  name: Analogy Questions (BATS)
42
  type: multiple-choice-qa
@@ -47,7 +47,7 @@ model-index:
47
  metrics:
48
  - name: Accuracy
49
  type: accuracy
50
- value: 0.783212896053363
51
  - task:
52
  name: Analogy Questions (Google)
53
  type: multiple-choice-qa
@@ -58,7 +58,7 @@ model-index:
58
  metrics:
59
  - name: Accuracy
60
  type: accuracy
61
- value: 0.934
62
  - task:
63
  name: Analogy Questions (U2)
64
  type: multiple-choice-qa
@@ -69,7 +69,7 @@ model-index:
69
  metrics:
70
  - name: Accuracy
71
  type: accuracy
72
- value: 0.6754385964912281
73
  - task:
74
  name: Analogy Questions (U4)
75
  type: multiple-choice-qa
@@ -80,7 +80,7 @@ model-index:
80
  metrics:
81
  - name: Accuracy
82
  type: accuracy
83
- value: 0.6388888888888888
84
  - task:
85
  name: Analogy Questions (ConceptNet Analogy)
86
  type: multiple-choice-qa
@@ -91,7 +91,7 @@ model-index:
91
  metrics:
92
  - name: Accuracy
93
  type: accuracy
94
- value: 0.43288590604026844
95
  - task:
96
  name: Analogy Questions (TREX Analogy)
97
  type: multiple-choice-qa
@@ -102,7 +102,7 @@ model-index:
102
  metrics:
103
  - name: Accuracy
104
  type: accuracy
105
- value: 0.6775956284153005
106
  - task:
107
  name: Analogy Questions (NELL-ONE Analogy)
108
  type: multiple-choice-qa
@@ -113,7 +113,7 @@ model-index:
113
  metrics:
114
  - name: Accuracy
115
  type: accuracy
116
- value: 0.605
117
  - task:
118
  name: Lexical Relation Classification (BLESS)
119
  type: classification
@@ -124,10 +124,10 @@ model-index:
124
  metrics:
125
  - name: F1
126
  type: f1
127
- value: 0.9148711767364774
128
  - name: F1 (macro)
129
  type: f1_macro
130
- value: 0.9119056356713013
131
  - task:
132
  name: Lexical Relation Classification (CogALexV)
133
  type: classification
@@ -138,10 +138,10 @@ model-index:
138
  metrics:
139
  - name: F1
140
  type: f1
141
- value: 0.8485915492957746
142
  - name: F1 (macro)
143
  type: f1_macro
144
- value: 0.6811794888962958
145
  - task:
146
  name: Lexical Relation Classification (EVALution)
147
  type: classification
@@ -152,10 +152,10 @@ model-index:
152
  metrics:
153
  - name: F1
154
  type: f1
155
- value: 0.6690140845070423
156
  - name: F1 (macro)
157
  type: f1_macro
158
- value: 0.6624009209291007
159
  - task:
160
  name: Lexical Relation Classification (K&H+N)
161
  type: classification
@@ -166,10 +166,10 @@ model-index:
166
  metrics:
167
  - name: F1
168
  type: f1
169
- value: 0.9508937886902692
170
  - name: F1 (macro)
171
  type: f1_macro
172
- value: 0.8677983904224069
173
  - task:
174
  name: Lexical Relation Classification (ROOT09)
175
  type: classification
@@ -180,34 +180,34 @@ model-index:
180
  metrics:
181
  - name: F1
182
  type: f1
183
- value: 0.8918834221247258
184
  - name: F1 (macro)
185
  type: f1_macro
186
- value: 0.8905814580868343
187
 
188
  ---
189
  # relbert/relbert-roberta-base-nce-semeval2012-0
190
 
191
- RelBERT based on [roberta-large](https://huggingface.co/roberta-large) fine-tuned on [relbert/semeval2012_relational_similarity](https://huggingface.co/datasets/relbert/semeval2012_relational_similarity) (see the [`relbert`](https://github.com/asahi417/relbert) for more detail of fine-tuning).
192
  This model achieves the following results on the relation understanding tasks:
193
  - Analogy Question ([dataset](https://huggingface.co/datasets/relbert/analogy_questions), [full result](https://huggingface.co/relbert/relbert-roberta-base-nce-semeval2012-0/raw/main/analogy.forward.json)):
194
- - Accuracy on SAT (full): 0.6818181818181818
195
- - Accuracy on SAT: 0.6824925816023739
196
- - Accuracy on BATS: 0.783212896053363
197
- - Accuracy on U2: 0.6754385964912281
198
- - Accuracy on U4: 0.6388888888888888
199
- - Accuracy on Google: 0.934
200
- - Accuracy on ConceptNet Analogy: 0.43288590604026844
201
- - Accuracy on T-Rex Analogy: 0.6775956284153005
202
- - Accuracy on NELL-ONE Analogy: 0.605
203
  - Lexical Relation Classification ([dataset](https://huggingface.co/datasets/relbert/lexical_relation_classification), [full result](https://huggingface.co/relbert/relbert-roberta-base-nce-semeval2012-0/raw/main/classification.json)):
204
- - Micro F1 score on BLESS: 0.9148711767364774
205
- - Micro F1 score on CogALexV: 0.8485915492957746
206
- - Micro F1 score on EVALution: 0.6690140845070423
207
- - Micro F1 score on K&H+N: 0.9508937886902692
208
- - Micro F1 score on ROOT09: 0.8918834221247258
209
  - Relation Mapping ([dataset](https://huggingface.co/datasets/relbert/relation_mapping), [full result](https://huggingface.co/relbert/relbert-roberta-base-nce-semeval2012-0/raw/main/relation_mapping.json)):
210
- - Accuracy on Relation Mapping: 0.8133333333333334
211
 
212
 
213
  ### Usage
@@ -224,7 +224,7 @@ vector = model.get_embedding(['Tokyo', 'Japan']) # shape of (n_dim, )
224
 
225
  ### Training hyperparameters
226
 
227
- - model: roberta-large
228
  - max_length: 64
229
  - epoch: 10
230
  - batch: 32
@@ -239,7 +239,7 @@ vector = model.get_embedding(['Tokyo', 'Japan']) # shape of (n_dim, )
239
  - split_valid: validation
240
  - loss_function: nce
241
  - classification_loss: False
242
- - loss_function_config: {'temperature': 0.05, 'num_negative': 100, 'num_positive': 10}
243
  - augment_negative_by_positive: True
244
 
245
  See the full configuration at [config file](https://huggingface.co/relbert/relbert-roberta-base-nce-semeval2012-0/raw/main/finetuning_config.json).
 
14
  metrics:
15
  - name: Accuracy
16
  type: accuracy
17
+ value: 0.817202380952381
18
  - task:
19
  name: Analogy Questions (SAT full)
20
  type: multiple-choice-qa
 
25
  metrics:
26
  - name: Accuracy
27
  type: accuracy
28
+ value: 0.5989304812834224
29
  - task:
30
  name: Analogy Questions (SAT)
31
  type: multiple-choice-qa
 
36
  metrics:
37
  - name: Accuracy
38
  type: accuracy
39
+ value: 0.6083086053412463
40
  - task:
41
  name: Analogy Questions (BATS)
42
  type: multiple-choice-qa
 
47
  metrics:
48
  - name: Accuracy
49
  type: accuracy
50
+ value: 0.7031684269038355
51
  - task:
52
  name: Analogy Questions (Google)
53
  type: multiple-choice-qa
 
58
  metrics:
59
  - name: Accuracy
60
  type: accuracy
61
+ value: 0.892
62
  - task:
63
  name: Analogy Questions (U2)
64
  type: multiple-choice-qa
 
69
  metrics:
70
  - name: Accuracy
71
  type: accuracy
72
+ value: 0.5964912280701754
73
  - task:
74
  name: Analogy Questions (U4)
75
  type: multiple-choice-qa
 
80
  metrics:
81
  - name: Accuracy
82
  type: accuracy
83
+ value: 0.5740740740740741
84
  - task:
85
  name: Analogy Questions (ConceptNet Analogy)
86
  type: multiple-choice-qa
 
91
  metrics:
92
  - name: Accuracy
93
  type: accuracy
94
+ value: 0.3976510067114094
95
  - task:
96
  name: Analogy Questions (TREX Analogy)
97
  type: multiple-choice-qa
 
102
  metrics:
103
  - name: Accuracy
104
  type: accuracy
105
+ value: 0.6666666666666666
106
  - task:
107
  name: Analogy Questions (NELL-ONE Analogy)
108
  type: multiple-choice-qa
 
113
  metrics:
114
  - name: Accuracy
115
  type: accuracy
116
+ value: 0.62
117
  - task:
118
  name: Lexical Relation Classification (BLESS)
119
  type: classification
 
124
  metrics:
125
  - name: F1
126
  type: f1
127
+ value: 0.8998041283712521
128
  - name: F1 (macro)
129
  type: f1_macro
130
+ value: 0.896201243435411
131
  - task:
132
  name: Lexical Relation Classification (CogALexV)
133
  type: classification
 
138
  metrics:
139
  - name: F1
140
  type: f1
141
+ value: 0.8370892018779342
142
  - name: F1 (macro)
143
  type: f1_macro
144
+ value: 0.6583174043371445
145
  - task:
146
  name: Lexical Relation Classification (EVALution)
147
  type: classification
 
152
  metrics:
153
  - name: F1
154
  type: f1
155
+ value: 0.6419284940411701
156
  - name: F1 (macro)
157
  type: f1_macro
158
+ value: 0.6294309369547718
159
  - task:
160
  name: Lexical Relation Classification (K&H+N)
161
  type: classification
 
166
  metrics:
167
  - name: F1
168
  type: f1
169
+ value: 0.9396953467343674
170
  - name: F1 (macro)
171
  type: f1_macro
172
+ value: 0.8459283973092365
173
  - task:
174
  name: Lexical Relation Classification (ROOT09)
175
  type: classification
 
180
  metrics:
181
  - name: F1
182
  type: f1
183
+ value: 0.8815418364149169
184
  - name: F1 (macro)
185
  type: f1_macro
186
+ value: 0.879329189992711
187
 
188
  ---
189
  # relbert/relbert-roberta-base-nce-semeval2012-0
190
 
191
+ RelBERT based on [roberta-base](https://huggingface.co/roberta-base) fine-tuned on [relbert/semeval2012_relational_similarity](https://huggingface.co/datasets/relbert/semeval2012_relational_similarity) (see the [`relbert`](https://github.com/asahi417/relbert) for more detail of fine-tuning).
192
  This model achieves the following results on the relation understanding tasks:
193
  - Analogy Question ([dataset](https://huggingface.co/datasets/relbert/analogy_questions), [full result](https://huggingface.co/relbert/relbert-roberta-base-nce-semeval2012-0/raw/main/analogy.forward.json)):
194
+ - Accuracy on SAT (full): 0.5989304812834224
195
+ - Accuracy on SAT: 0.6083086053412463
196
+ - Accuracy on BATS: 0.7031684269038355
197
+ - Accuracy on U2: 0.5964912280701754
198
+ - Accuracy on U4: 0.5740740740740741
199
+ - Accuracy on Google: 0.892
200
+ - Accuracy on ConceptNet Analogy: 0.3976510067114094
201
+ - Accuracy on T-Rex Analogy: 0.6666666666666666
202
+ - Accuracy on NELL-ONE Analogy: 0.62
203
  - Lexical Relation Classification ([dataset](https://huggingface.co/datasets/relbert/lexical_relation_classification), [full result](https://huggingface.co/relbert/relbert-roberta-base-nce-semeval2012-0/raw/main/classification.json)):
204
+ - Micro F1 score on BLESS: 0.8998041283712521
205
+ - Micro F1 score on CogALexV: 0.8370892018779342
206
+ - Micro F1 score on EVALution: 0.6419284940411701
207
+ - Micro F1 score on K&H+N: 0.9396953467343674
208
+ - Micro F1 score on ROOT09: 0.8815418364149169
209
  - Relation Mapping ([dataset](https://huggingface.co/datasets/relbert/relation_mapping), [full result](https://huggingface.co/relbert/relbert-roberta-base-nce-semeval2012-0/raw/main/relation_mapping.json)):
210
+ - Accuracy on Relation Mapping: 0.817202380952381
211
 
212
 
213
  ### Usage
 
224
 
225
  ### Training hyperparameters
226
 
227
+ - model: roberta-base
228
  - max_length: 64
229
  - epoch: 10
230
  - batch: 32
 
239
  - split_valid: validation
240
  - loss_function: nce
241
  - classification_loss: False
242
+ - loss_function_config: {'temperature': 0.05, 'num_negative': 400, 'num_positive': 10}
243
  - augment_negative_by_positive: True
244
 
245
  See the full configuration at [config file](https://huggingface.co/relbert/relbert-roberta-base-nce-semeval2012-0/raw/main/finetuning_config.json).
analogy.forward.json CHANGED
@@ -1 +1 @@
1
- {"semeval2012_relational_similarity/validation": 0.7468354430379747, "scan/test": 0.2592821782178218, "sat_full/test": 0.6818181818181818, "sat/test": 0.6824925816023739, "u2/test": 0.6754385964912281, "u4/test": 0.6388888888888888, "google/test": 0.934, "bats/test": 0.783212896053363, "t_rex_relational_similarity/test": 0.6775956284153005, "conceptnet_relational_similarity/test": 0.43288590604026844, "nell_relational_similarity/test": 0.605}
 
1
+ {"semeval2012_relational_similarity/validation": 0.7848101265822784, "scan/test": 0.2592821782178218, "sat_full/test": 0.5989304812834224, "sat/test": 0.6083086053412463, "u2/test": 0.5964912280701754, "u4/test": 0.5740740740740741, "google/test": 0.892, "bats/test": 0.7031684269038355, "t_rex_relational_similarity/test": 0.6666666666666666, "conceptnet_relational_similarity/test": 0.3976510067114094, "nell_relational_similarity/test": 0.62, "scan/validation": 0.25842696629213485, "sat/validation": 0.5135135135135135, "u2/validation": 0.4583333333333333, "u4/validation": 0.6458333333333334, "google/validation": 0.96, "bats/validation": 0.7738693467336684, "t_rex_relational_similarity/validation": 0.2661290322580645, "conceptnet_relational_similarity/validation": 0.32823741007194246, "nell_relational_similarity/validation": 0.575}
classification.json CHANGED
@@ -1 +1 @@
1
- {"lexical_relation_classification/BLESS": {"classifier_config": {"activation": "relu", "alpha": 0.0001, "batch_size": "auto", "beta_1": 0.9, "beta_2": 0.999, "early_stopping": false, "epsilon": 1e-08, "hidden_layer_sizes": [100], "learning_rate": "constant", "learning_rate_init": 0.001, "max_fun": 15000, "max_iter": 200, "momentum": 0.9, "n_iter_no_change": 10, "nesterovs_momentum": true, "power_t": 0.5, "random_state": 0, "shuffle": true, "solver": "adam", "tol": 0.0001, "validation_fraction": 0.1, "verbose": false, "warm_start": false}, "test/accuracy": 0.9148711767364773, "test/f1_macro": 0.9119056356713013, "test/f1_micro": 0.9148711767364774, "test/p_macro": 0.908504306097206, "test/p_micro": 0.9148711767364773, "test/r_macro": 0.9162009423798682, "test/r_micro": 0.9148711767364773, "test/f1/attri": 0.9090909090909092, "test/p/attri": 0.9130434782608695, "test/r/attri": 0.9051724137931034, "test/f1/coord": 0.9614309670206819, "test/p/coord": 0.948180815876516, "test/r/coord": 0.9750566893424036, "test/f1/event": 0.8515742128935532, "test/p/event": 0.8145315487571702, "test/r/event": 0.8921465968586387, "test/f1/hyper": 0.9331395348837209, "test/p/hyper": 0.9497041420118343, "test/r/hyper": 0.9171428571428571, "test/f1/mero": 0.8878566688785667, "test/p/mero": 0.8791064388961892, "test/r/mero": 0.8967828418230563, "test/f1/random": 0.9283415212603762, "test/p/random": 0.9464594127806563, "test/r/random": 0.910904255319149}, "lexical_relation_classification/CogALexV": {"classifier_config": {"activation": "relu", "alpha": 0.0001, "batch_size": "auto", "beta_1": 0.9, "beta_2": 0.999, "early_stopping": false, "epsilon": 1e-08, "hidden_layer_sizes": [100], "learning_rate": "constant", "learning_rate_init": 0.001, "max_fun": 15000, "max_iter": 200, "momentum": 0.9, "n_iter_no_change": 10, "nesterovs_momentum": true, "power_t": 0.5, "random_state": 0, "shuffle": true, "solver": "adam", "tol": 0.0001, "validation_fraction": 0.1, "verbose": false, "warm_start": false}, "test/accuracy": 0.8485915492957746, "test/f1_macro": 0.6811794888962958, "test/f1_micro": 0.8485915492957746, "test/p_macro": 0.716549900315324, "test/p_micro": 0.8485915492957746, "test/r_macro": 0.6520962513564919, "test/r_micro": 0.8485915492957746, "test/f1/ANT": 0.7326440177252584, "test/p/ANT": 0.7823343848580442, "test/r/ANT": 0.6888888888888889, "test/f1/HYPER": 0.6017441860465116, "test/p/HYPER": 0.6764705882352942, "test/r/HYPER": 0.5418848167539267, "test/f1/PART_OF": 0.6872037914691943, "test/p/PART_OF": 0.7323232323232324, "test/r/PART_OF": 0.6473214285714286, "test/f1/RANDOM": 0.926255562619199, "test/p/RANDOM": 0.9013300340241261, "test/r/RANDOM": 0.952598888525662, "test/f1/SYN": 0.4580498866213152, "test/p/SYN": 0.49029126213592233, "test/r/SYN": 0.4297872340425532}, "lexical_relation_classification/EVALution": {"classifier_config": {"activation": "relu", "alpha": 0.0001, "batch_size": "auto", "beta_1": 0.9, "beta_2": 0.999, "early_stopping": false, "epsilon": 1e-08, "hidden_layer_sizes": [100], "learning_rate": "constant", "learning_rate_init": 0.001, "max_fun": 15000, "max_iter": 200, "momentum": 0.9, "n_iter_no_change": 10, "nesterovs_momentum": true, "power_t": 0.5, "random_state": 0, "shuffle": true, "solver": "adam", "tol": 0.0001, "validation_fraction": 0.1, "verbose": false, "warm_start": false}, "test/accuracy": 0.6690140845070423, "test/f1_macro": 0.6624009209291007, "test/f1_micro": 0.6690140845070423, "test/p_macro": 0.6762790775482931, "test/p_micro": 0.6690140845070423, "test/r_macro": 0.6547629962331695, "test/r_micro": 0.6690140845070423, "test/f1/Antonym": 0.7820512820512822, "test/p/Antonym": 0.8356164383561644, "test/r/Antonym": 0.7349397590361446, "test/f1/HasA": 0.6754098360655738, "test/p/HasA": 0.6319018404907976, "test/r/HasA": 0.7253521126760564, "test/f1/HasProperty": 0.8161434977578476, "test/p/HasProperty": 0.7867435158501441, "test/r/HasProperty": 0.8478260869565217, "test/f1/IsA": 0.6150670794633643, "test/p/IsA": 0.5843137254901961, "test/r/IsA": 0.6492374727668845, "test/f1/MadeOf": 0.6282051282051283, "test/p/MadeOf": 0.7, "test/r/MadeOf": 0.5697674418604651, "test/f1/PartOf": 0.706766917293233, "test/p/PartOf": 0.7768595041322314, "test/r/PartOf": 0.6482758620689655, "test/f1/Synonym": 0.41316270566727603, "test/p/Synonym": 0.4185185185185185, "test/r/Synonym": 0.40794223826714804}, "lexical_relation_classification/K&H+N": {"classifier_config": {"activation": "relu", "alpha": 0.0001, "batch_size": "auto", "beta_1": 0.9, "beta_2": 0.999, "early_stopping": false, "epsilon": 1e-08, "hidden_layer_sizes": [100], "learning_rate": "constant", "learning_rate_init": 0.001, "max_fun": 15000, "max_iter": 200, "momentum": 0.9, "n_iter_no_change": 10, "nesterovs_momentum": true, "power_t": 0.5, "random_state": 0, "shuffle": true, "solver": "adam", "tol": 0.0001, "validation_fraction": 0.1, "verbose": false, "warm_start": false}, "test/accuracy": 0.9508937886902692, "test/f1_macro": 0.8677983904224069, "test/f1_micro": 0.9508937886902692, "test/p_macro": 0.8771720388507485, "test/p_micro": 0.9508937886902692, "test/r_macro": 0.8593517976091283, "test/r_micro": 0.9508937886902692, "test/f1/false": 0.960130428338521, "test/p/false": 0.9599881446354476, "test/r/false": 0.9602727542247258, "test/f1/hypo": 0.9278557114228457, "test/p/hypo": 0.9706498951781971, "test/r/hypo": 0.8886756238003839, "test/f1/mero": 0.6263048016701461, "test/p/mero": 0.6276150627615062, "test/r/mero": 0.625, "test/f1/sibl": 0.956902620258115, "test/p/sibl": 0.9504350528278434, "test/r/sibl": 0.9634588124114034}, "lexical_relation_classification/ROOT09": {"classifier_config": {"activation": "relu", "alpha": 0.0001, "batch_size": "auto", "beta_1": 0.9, "beta_2": 0.999, "early_stopping": false, "epsilon": 1e-08, "hidden_layer_sizes": [100], "learning_rate": "constant", "learning_rate_init": 0.001, "max_fun": 15000, "max_iter": 200, "momentum": 0.9, "n_iter_no_change": 10, "nesterovs_momentum": true, "power_t": 0.5, "random_state": 0, "shuffle": true, "solver": "adam", "tol": 0.0001, "validation_fraction": 0.1, "verbose": false, "warm_start": false}, "test/accuracy": 0.8918834221247258, "test/f1_macro": 0.8905814580868343, "test/f1_micro": 0.8918834221247258, "test/p_macro": 0.8915743175976042, "test/p_micro": 0.8918834221247258, "test/r_macro": 0.8896098856499345, "test/r_micro": 0.8918834221247258, "test/f1/COORD": 0.971709717097171, "test/p/COORD": 0.9753086419753086, "test/r/COORD": 0.9681372549019608, "test/f1/HYPER": 0.8047117172969622, "test/p/HYPER": 0.8072139303482587, "test/r/HYPER": 0.8022249690976514, "test/f1/RANDOM": 0.8953229398663697, "test/p/RANDOM": 0.8922003804692454, "test/r/RANDOM": 0.8984674329501916}}
 
1
+ {"lexical_relation_classification/BLESS": {"classifier_config": {"activation": "relu", "alpha": 0.0001, "batch_size": "auto", "beta_1": 0.9, "beta_2": 0.999, "early_stopping": false, "epsilon": 1e-08, "hidden_layer_sizes": [100], "learning_rate": "constant", "learning_rate_init": 0.001, "max_fun": 15000, "max_iter": 200, "momentum": 0.9, "n_iter_no_change": 10, "nesterovs_momentum": true, "power_t": 0.5, "random_state": 0, "shuffle": true, "solver": "adam", "tol": 0.0001, "validation_fraction": 0.1, "verbose": false, "warm_start": false}, "test/accuracy": 0.8998041283712521, "test/f1_macro": 0.896201243435411, "test/f1_micro": 0.8998041283712521, "test/p_macro": 0.8876829436591316, "test/p_micro": 0.8998041283712521, "test/r_macro": 0.9054007585142311, "test/r_micro": 0.8998041283712521}, "lexical_relation_classification/CogALexV": {"classifier_config": {"activation": "relu", "alpha": 0.0001, "batch_size": "auto", "beta_1": 0.9, "beta_2": 0.999, "early_stopping": false, "epsilon": 1e-08, "hidden_layer_sizes": [100], "learning_rate": "constant", "learning_rate_init": 0.001, "max_fun": 15000, "max_iter": 200, "momentum": 0.9, "n_iter_no_change": 10, "nesterovs_momentum": true, "power_t": 0.5, "random_state": 0, "shuffle": true, "solver": "adam", "tol": 0.0001, "validation_fraction": 0.1, "verbose": false, "warm_start": false}, "test/accuracy": 0.8370892018779342, "test/f1_macro": 0.6583174043371445, "test/f1_micro": 0.8370892018779342, "test/p_macro": 0.6822907887970884, "test/p_micro": 0.8370892018779342, "test/r_macro": 0.6384370436284232, "test/r_micro": 0.8370892018779342}, "lexical_relation_classification/EVALution": {"classifier_config": {"activation": "relu", "alpha": 0.0001, "batch_size": "auto", "beta_1": 0.9, "beta_2": 0.999, "early_stopping": false, "epsilon": 1e-08, "hidden_layer_sizes": [100], "learning_rate": "constant", "learning_rate_init": 0.001, "max_fun": 15000, "max_iter": 200, "momentum": 0.9, "n_iter_no_change": 10, "nesterovs_momentum": true, "power_t": 0.5, "random_state": 0, "shuffle": true, "solver": "adam", "tol": 0.0001, "validation_fraction": 0.1, "verbose": false, "warm_start": false}, "test/accuracy": 0.6419284940411701, "test/f1_macro": 0.6294309369547718, "test/f1_micro": 0.6419284940411701, "test/p_macro": 0.6360186480100325, "test/p_micro": 0.6419284940411701, "test/r_macro": 0.6300178037199379, "test/r_micro": 0.6419284940411701}, "lexical_relation_classification/K&H+N": {"classifier_config": {"activation": "relu", "alpha": 0.0001, "batch_size": "auto", "beta_1": 0.9, "beta_2": 0.999, "early_stopping": false, "epsilon": 1e-08, "hidden_layer_sizes": [100], "learning_rate": "constant", "learning_rate_init": 0.001, "max_fun": 15000, "max_iter": 200, "momentum": 0.9, "n_iter_no_change": 10, "nesterovs_momentum": true, "power_t": 0.5, "random_state": 0, "shuffle": true, "solver": "adam", "tol": 0.0001, "validation_fraction": 0.1, "verbose": false, "warm_start": false}, "test/accuracy": 0.9396953467343674, "test/f1_macro": 0.8459283973092365, "test/f1_micro": 0.9396953467343674, "test/p_macro": 0.8614600859106621, "test/p_micro": 0.9396953467343674, "test/r_macro": 0.8351465630922283, "test/r_micro": 0.9396953467343674}, "lexical_relation_classification/ROOT09": {"classifier_config": {"activation": "relu", "alpha": 0.0001, "batch_size": "auto", "beta_1": 0.9, "beta_2": 0.999, "early_stopping": false, "epsilon": 1e-08, "hidden_layer_sizes": [100], "learning_rate": "constant", "learning_rate_init": 0.001, "max_fun": 15000, "max_iter": 200, "momentum": 0.9, "n_iter_no_change": 10, "nesterovs_momentum": true, "power_t": 0.5, "random_state": 0, "shuffle": true, "solver": "adam", "tol": 0.0001, "validation_fraction": 0.1, "verbose": false, "warm_start": false}, "test/accuracy": 0.8815418364149169, "test/f1_macro": 0.879329189992711, "test/f1_micro": 0.8815418364149169, "test/p_macro": 0.8763389203201842, "test/p_micro": 0.8815418364149169, "test/r_macro": 0.882560877928503, "test/r_micro": 0.8815418364149169}}
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "roberta-large",
3
  "architectures": [
4
  "RobertaModel"
5
  ],
@@ -9,14 +9,14 @@
9
  "eos_token_id": 2,
10
  "hidden_act": "gelu",
11
  "hidden_dropout_prob": 0.1,
12
- "hidden_size": 1024,
13
  "initializer_range": 0.02,
14
- "intermediate_size": 4096,
15
  "layer_norm_eps": 1e-05,
16
  "max_position_embeddings": 514,
17
  "model_type": "roberta",
18
- "num_attention_heads": 16,
19
- "num_hidden_layers": 24,
20
  "pad_token_id": 1,
21
  "position_embedding_type": "absolute",
22
  "relbert_config": {
@@ -24,7 +24,7 @@
24
  "template": "Today, I finally discovered the relation between <subj> and <obj> : <subj> is the <mask> of <obj>"
25
  },
26
  "torch_dtype": "float32",
27
- "transformers_version": "4.26.1",
28
  "type_vocab_size": 1,
29
  "use_cache": true,
30
  "vocab_size": 50265
 
1
  {
2
+ "_name_or_path": "roberta-base",
3
  "architectures": [
4
  "RobertaModel"
5
  ],
 
9
  "eos_token_id": 2,
10
  "hidden_act": "gelu",
11
  "hidden_dropout_prob": 0.1,
12
+ "hidden_size": 768,
13
  "initializer_range": 0.02,
14
+ "intermediate_size": 3072,
15
  "layer_norm_eps": 1e-05,
16
  "max_position_embeddings": 514,
17
  "model_type": "roberta",
18
+ "num_attention_heads": 12,
19
+ "num_hidden_layers": 12,
20
  "pad_token_id": 1,
21
  "position_embedding_type": "absolute",
22
  "relbert_config": {
 
24
  "template": "Today, I finally discovered the relation between <subj> and <obj> : <subj> is the <mask> of <obj>"
25
  },
26
  "torch_dtype": "float32",
27
+ "transformers_version": "4.16.2",
28
  "type_vocab_size": 1,
29
  "use_cache": true,
30
  "vocab_size": 50265
finetuning_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "template": "Today, I finally discovered the relation between <subj> and <obj> : <subj> is the <mask> of <obj>",
3
- "model": "roberta-large",
4
  "max_length": 64,
5
  "epoch": 10,
6
  "batch": 32,
@@ -17,7 +17,7 @@
17
  "classification_loss": false,
18
  "loss_function_config": {
19
  "temperature": 0.05,
20
- "num_negative": 100,
21
  "num_positive": 10
22
  },
23
  "augment_negative_by_positive": true
 
1
  {
2
  "template": "Today, I finally discovered the relation between <subj> and <obj> : <subj> is the <mask> of <obj>",
3
+ "model": "roberta-base",
4
  "max_length": 64,
5
  "epoch": 10,
6
  "batch": 32,
 
17
  "classification_loss": false,
18
  "loss_function_config": {
19
  "temperature": 0.05,
20
+ "num_negative": 400,
21
  "num_positive": 10
22
  },
23
  "augment_negative_by_positive": true
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3222067087f440c0c63e7bec1c581c12ba961c9c47a777787f99bad40f363360
3
- size 1421575277
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c94a92b29ec3435777fd39e26c2ad122e32bb0d45e73dc050bca78cbc525fd26
3
+ size 498663921
relation_mapping.json CHANGED
The diff for this file is too large to render. See raw diff
 
special_tokens_map.json CHANGED
@@ -1,15 +1 @@
1
- {
2
- "bos_token": "<s>",
3
- "cls_token": "<s>",
4
- "eos_token": "</s>",
5
- "mask_token": {
6
- "content": "<mask>",
7
- "lstrip": true,
8
- "normalized": false,
9
- "rstrip": false,
10
- "single_word": false
11
- },
12
- "pad_token": "<pad>",
13
- "sep_token": "</s>",
14
- "unk_token": "<unk>"
15
- }
 
1
+ {"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": false}}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
tokenizer.json CHANGED
@@ -53,8 +53,7 @@
53
  "pre_tokenizer": {
54
  "type": "ByteLevel",
55
  "add_prefix_space": false,
56
- "trim_offsets": true,
57
- "use_regex": true
58
  },
59
  "post_processor": {
60
  "type": "RobertaProcessing",
@@ -72,8 +71,7 @@
72
  "decoder": {
73
  "type": "ByteLevel",
74
  "add_prefix_space": true,
75
- "trim_offsets": true,
76
- "use_regex": true
77
  },
78
  "model": {
79
  "type": "BPE",
 
53
  "pre_tokenizer": {
54
  "type": "ByteLevel",
55
  "add_prefix_space": false,
56
+ "trim_offsets": true
 
57
  },
58
  "post_processor": {
59
  "type": "RobertaProcessing",
 
71
  "decoder": {
72
  "type": "ByteLevel",
73
  "add_prefix_space": true,
74
+ "trim_offsets": true
 
75
  },
76
  "model": {
77
  "type": "BPE",
tokenizer_config.json CHANGED
@@ -1,16 +1 @@
1
- {
2
- "add_prefix_space": false,
3
- "bos_token": "<s>",
4
- "cls_token": "<s>",
5
- "eos_token": "</s>",
6
- "errors": "replace",
7
- "mask_token": "<mask>",
8
- "model_max_length": 512,
9
- "name_or_path": "roberta-large",
10
- "pad_token": "<pad>",
11
- "sep_token": "</s>",
12
- "special_tokens_map_file": null,
13
- "tokenizer_class": "RobertaTokenizer",
14
- "trim_offsets": true,
15
- "unk_token": "<unk>"
16
- }
 
1
+ {"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "add_prefix_space": false, "errors": "replace", "sep_token": "</s>", "cls_token": "<s>", "pad_token": "<pad>", "mask_token": "<mask>", "trim_offsets": true, "model_max_length": 512, "special_tokens_map_file": null, "name_or_path": "roberta-base", "tokenizer_class": "RobertaTokenizer"}