model update
Browse files- README.md +268 -0
- analogy.bidirection.json +1 -0
- analogy.forward.json +1 -0
- analogy.reverse.json +1 -0
- classification.json +1 -0
- config.json +1 -1
- finetuning_config.json +24 -0
- relation_mapping.json +0 -0
- tokenizer_config.json +1 -1
README.md
ADDED
@@ -0,0 +1,268 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
datasets:
|
3 |
+
- relbert/semeval2012_relational_similarity
|
4 |
+
model-index:
|
5 |
+
- name: relbert/relbert-roberta-large-iloob-d-semeval2012
|
6 |
+
results:
|
7 |
+
- task:
|
8 |
+
name: Relation Mapping
|
9 |
+
type: sorting-task
|
10 |
+
dataset:
|
11 |
+
name: Relation Mapping
|
12 |
+
args: relbert/relation_mapping
|
13 |
+
type: relation-mapping
|
14 |
+
metrics:
|
15 |
+
- name: Accuracy
|
16 |
+
type: accuracy
|
17 |
+
value: 0.844484126984127
|
18 |
+
- task:
|
19 |
+
name: Analogy Questions (SAT full)
|
20 |
+
type: multiple-choice-qa
|
21 |
+
dataset:
|
22 |
+
name: SAT full
|
23 |
+
args: relbert/analogy_questions
|
24 |
+
type: analogy-questions
|
25 |
+
metrics:
|
26 |
+
- name: Accuracy
|
27 |
+
type: accuracy
|
28 |
+
value: 0.6764705882352942
|
29 |
+
- task:
|
30 |
+
name: Analogy Questions (SAT)
|
31 |
+
type: multiple-choice-qa
|
32 |
+
dataset:
|
33 |
+
name: SAT
|
34 |
+
args: relbert/analogy_questions
|
35 |
+
type: analogy-questions
|
36 |
+
metrics:
|
37 |
+
- name: Accuracy
|
38 |
+
type: accuracy
|
39 |
+
value: 0.6735905044510386
|
40 |
+
- task:
|
41 |
+
name: Analogy Questions (BATS)
|
42 |
+
type: multiple-choice-qa
|
43 |
+
dataset:
|
44 |
+
name: BATS
|
45 |
+
args: relbert/analogy_questions
|
46 |
+
type: analogy-questions
|
47 |
+
metrics:
|
48 |
+
- name: Accuracy
|
49 |
+
type: accuracy
|
50 |
+
value: 0.7926625903279599
|
51 |
+
- task:
|
52 |
+
name: Analogy Questions (Google)
|
53 |
+
type: multiple-choice-qa
|
54 |
+
dataset:
|
55 |
+
name: Google
|
56 |
+
args: relbert/analogy_questions
|
57 |
+
type: analogy-questions
|
58 |
+
metrics:
|
59 |
+
- name: Accuracy
|
60 |
+
type: accuracy
|
61 |
+
value: 0.968
|
62 |
+
- task:
|
63 |
+
name: Analogy Questions (U2)
|
64 |
+
type: multiple-choice-qa
|
65 |
+
dataset:
|
66 |
+
name: U2
|
67 |
+
args: relbert/analogy_questions
|
68 |
+
type: analogy-questions
|
69 |
+
metrics:
|
70 |
+
- name: Accuracy
|
71 |
+
type: accuracy
|
72 |
+
value: 0.6052631578947368
|
73 |
+
- task:
|
74 |
+
name: Analogy Questions (U4)
|
75 |
+
type: multiple-choice-qa
|
76 |
+
dataset:
|
77 |
+
name: U4
|
78 |
+
args: relbert/analogy_questions
|
79 |
+
type: analogy-questions
|
80 |
+
metrics:
|
81 |
+
- name: Accuracy
|
82 |
+
type: accuracy
|
83 |
+
value: 0.6226851851851852
|
84 |
+
- task:
|
85 |
+
name: Analogy Questions (ConceptNet Analogy)
|
86 |
+
type: multiple-choice-qa
|
87 |
+
dataset:
|
88 |
+
name: ConceptNet Analogy
|
89 |
+
args: relbert/analogy_questions
|
90 |
+
type: analogy-questions
|
91 |
+
metrics:
|
92 |
+
- name: Accuracy
|
93 |
+
type: accuracy
|
94 |
+
value: 0.39093959731543626
|
95 |
+
- task:
|
96 |
+
name: Analogy Questions (TREX Analogy)
|
97 |
+
type: multiple-choice-qa
|
98 |
+
dataset:
|
99 |
+
name: TREX Analogy
|
100 |
+
args: relbert/analogy_questions
|
101 |
+
type: analogy-questions
|
102 |
+
metrics:
|
103 |
+
- name: Accuracy
|
104 |
+
type: accuracy
|
105 |
+
value: 0.6612021857923497
|
106 |
+
- task:
|
107 |
+
name: Analogy Questions (NELL-ONE Analogy)
|
108 |
+
type: multiple-choice-qa
|
109 |
+
dataset:
|
110 |
+
name: NELL-ONE Analogy
|
111 |
+
args: relbert/analogy_questions
|
112 |
+
type: analogy-questions
|
113 |
+
metrics:
|
114 |
+
- name: Accuracy
|
115 |
+
type: accuracy
|
116 |
+
value: 0.6416666666666667
|
117 |
+
- task:
|
118 |
+
name: Lexical Relation Classification (BLESS)
|
119 |
+
type: classification
|
120 |
+
dataset:
|
121 |
+
name: BLESS
|
122 |
+
args: relbert/lexical_relation_classification
|
123 |
+
type: relation-classification
|
124 |
+
metrics:
|
125 |
+
- name: F1
|
126 |
+
type: f1
|
127 |
+
value: 0.9171312339912611
|
128 |
+
- name: F1 (macro)
|
129 |
+
type: f1_macro
|
130 |
+
value: 0.9144771714929822
|
131 |
+
- task:
|
132 |
+
name: Lexical Relation Classification (CogALexV)
|
133 |
+
type: classification
|
134 |
+
dataset:
|
135 |
+
name: CogALexV
|
136 |
+
args: relbert/lexical_relation_classification
|
137 |
+
type: relation-classification
|
138 |
+
metrics:
|
139 |
+
- name: F1
|
140 |
+
type: f1
|
141 |
+
value: 0.8455399061032864
|
142 |
+
- name: F1 (macro)
|
143 |
+
type: f1_macro
|
144 |
+
value: 0.6723727069325071
|
145 |
+
- task:
|
146 |
+
name: Lexical Relation Classification (EVALution)
|
147 |
+
type: classification
|
148 |
+
dataset:
|
149 |
+
name: BLESS
|
150 |
+
args: relbert/lexical_relation_classification
|
151 |
+
type: relation-classification
|
152 |
+
metrics:
|
153 |
+
- name: F1
|
154 |
+
type: f1
|
155 |
+
value: 0.6793066088840737
|
156 |
+
- name: F1 (macro)
|
157 |
+
type: f1_macro
|
158 |
+
value: 0.6676508193568021
|
159 |
+
- task:
|
160 |
+
name: Lexical Relation Classification (K&H+N)
|
161 |
+
type: classification
|
162 |
+
dataset:
|
163 |
+
name: K&H+N
|
164 |
+
args: relbert/lexical_relation_classification
|
165 |
+
type: relation-classification
|
166 |
+
metrics:
|
167 |
+
- name: F1
|
168 |
+
type: f1
|
169 |
+
value: 0.95875356472143
|
170 |
+
- name: F1 (macro)
|
171 |
+
type: f1_macro
|
172 |
+
value: 0.8812996852310372
|
173 |
+
- task:
|
174 |
+
name: Lexical Relation Classification (ROOT09)
|
175 |
+
type: classification
|
176 |
+
dataset:
|
177 |
+
name: ROOT09
|
178 |
+
args: relbert/lexical_relation_classification
|
179 |
+
type: relation-classification
|
180 |
+
metrics:
|
181 |
+
- name: F1
|
182 |
+
type: f1
|
183 |
+
value: 0.9072391099968662
|
184 |
+
- name: F1 (macro)
|
185 |
+
type: f1_macro
|
186 |
+
value: 0.9041555105317078
|
187 |
+
|
188 |
+
---
|
189 |
+
# relbert/relbert-roberta-large-iloob-d-semeval2012
|
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-large-iloob-d-semeval2012/raw/main/analogy.forward.json)):
|
194 |
+
- Accuracy on SAT (full): 0.6764705882352942
|
195 |
+
- Accuracy on SAT: 0.6735905044510386
|
196 |
+
- Accuracy on BATS: 0.7926625903279599
|
197 |
+
- Accuracy on U2: 0.6052631578947368
|
198 |
+
- Accuracy on U4: 0.6226851851851852
|
199 |
+
- Accuracy on Google: 0.968
|
200 |
+
- Accuracy on ConceptNet Analogy: 0.39093959731543626
|
201 |
+
- Accuracy on T-Rex Analogy: 0.6612021857923497
|
202 |
+
- Accuracy on NELL-ONE Analogy: 0.6416666666666667
|
203 |
+
- Lexical Relation Classification ([dataset](https://huggingface.co/datasets/relbert/lexical_relation_classification), [full result](https://huggingface.co/relbert/relbert-roberta-large-iloob-d-semeval2012/raw/main/classification.json)):
|
204 |
+
- Micro F1 score on BLESS: 0.9171312339912611
|
205 |
+
- Micro F1 score on CogALexV: 0.8455399061032864
|
206 |
+
- Micro F1 score on EVALution: 0.6793066088840737
|
207 |
+
- Micro F1 score on K&H+N: 0.95875356472143
|
208 |
+
- Micro F1 score on ROOT09: 0.9072391099968662
|
209 |
+
- Relation Mapping ([dataset](https://huggingface.co/datasets/relbert/relation_mapping), [full result](https://huggingface.co/relbert/relbert-roberta-large-iloob-d-semeval2012/raw/main/relation_mapping.json)):
|
210 |
+
- Accuracy on Relation Mapping: 0.844484126984127
|
211 |
+
|
212 |
+
|
213 |
+
### Usage
|
214 |
+
This model can be used through the [relbert library](https://github.com/asahi417/relbert). Install the library via pip
|
215 |
+
```shell
|
216 |
+
pip install relbert
|
217 |
+
```
|
218 |
+
and activate model as below.
|
219 |
+
```python
|
220 |
+
from relbert import RelBERT
|
221 |
+
model = RelBERT("relbert/relbert-roberta-large-iloob-d-semeval2012")
|
222 |
+
vector = model.get_embedding(['Tokyo', 'Japan']) # shape of (n_dim, )
|
223 |
+
```
|
224 |
+
|
225 |
+
### Training hyperparameters
|
226 |
+
|
227 |
+
- model: roberta-large
|
228 |
+
- max_length: 64
|
229 |
+
- epoch: 10
|
230 |
+
- batch: 32
|
231 |
+
- random_seed: 0
|
232 |
+
- lr: 5e-06
|
233 |
+
- lr_warmup: 10
|
234 |
+
- aggregation_mode: average_no_mask
|
235 |
+
- data: relbert/semeval2012_relational_similarity
|
236 |
+
- data_name: None
|
237 |
+
- exclude_relation: None
|
238 |
+
- split: train
|
239 |
+
- split_valid: validation
|
240 |
+
- loss_function: iloob
|
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-large-iloob-d-semeval2012/raw/main/finetuning_config.json).
|
246 |
+
|
247 |
+
### Reference
|
248 |
+
If you use any resource from RelBERT, please consider to cite our [paper](https://aclanthology.org/2021.emnlp-main.712/).
|
249 |
+
|
250 |
+
```
|
251 |
+
|
252 |
+
@inproceedings{ushio-etal-2021-distilling,
|
253 |
+
title = "Distilling Relation Embeddings from Pretrained Language Models",
|
254 |
+
author = "Ushio, Asahi and
|
255 |
+
Camacho-Collados, Jose and
|
256 |
+
Schockaert, Steven",
|
257 |
+
booktitle = "Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing",
|
258 |
+
month = nov,
|
259 |
+
year = "2021",
|
260 |
+
address = "Online and Punta Cana, Dominican Republic",
|
261 |
+
publisher = "Association for Computational Linguistics",
|
262 |
+
url = "https://aclanthology.org/2021.emnlp-main.712",
|
263 |
+
doi = "10.18653/v1/2021.emnlp-main.712",
|
264 |
+
pages = "9044--9062",
|
265 |
+
abstract = "Pre-trained language models have been found to capture a surprisingly rich amount of lexical knowledge, ranging from commonsense properties of everyday concepts to detailed factual knowledge about named entities. Among others, this makes it possible to distill high-quality word vectors from pre-trained language models. However, it is currently unclear to what extent it is possible to distill relation embeddings, i.e. vectors that characterize the relationship between two words. Such relation embeddings are appealing because they can, in principle, encode relational knowledge in a more fine-grained way than is possible with knowledge graphs. To obtain relation embeddings from a pre-trained language model, we encode word pairs using a (manually or automatically generated) prompt, and we fine-tune the language model such that relationally similar word pairs yield similar output vectors. We find that the resulting relation embeddings are highly competitive on analogy (unsupervised) and relation classification (supervised) benchmarks, even without any task-specific fine-tuning. Source code to reproduce our experimental results and the model checkpoints are available in the following repository: https://github.com/asahi417/relbert",
|
266 |
+
}
|
267 |
+
|
268 |
+
```
|
analogy.bidirection.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"sat_full/test": 0.7085561497326203, "sat/test": 0.7091988130563798, "u2/test": 0.6666666666666666, "u4/test": 0.6666666666666666, "google/test": 0.978, "bats/test": 0.8293496386881601, "t_rex_relational_similarity/test": 0.6612021857923497, "conceptnet_relational_similarity/test": 0.4236577181208054, "nell_relational_similarity/test": 0.735, "sat/validation": 0.7027027027027027, "u2/validation": 0.625, "u4/validation": 0.6458333333333334, "google/validation": 1.0, "bats/validation": 0.8793969849246231, "semeval2012_relational_similarity/validation": 0.7215189873417721, "t_rex_relational_similarity/validation": 0.3165322580645161, "conceptnet_relational_similarity/validation": 0.3597122302158273, "nell_relational_similarity/validation": 0.6475}
|
analogy.forward.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"semeval2012_relational_similarity/validation": 0.759493670886076, "sat_full/test": 0.6764705882352942, "sat/test": 0.6735905044510386, "u2/test": 0.6052631578947368, "u4/test": 0.6226851851851852, "google/test": 0.968, "bats/test": 0.7926625903279599, "t_rex_relational_similarity/test": 0.6612021857923497, "conceptnet_relational_similarity/test": 0.39093959731543626, "nell_relational_similarity/test": 0.6416666666666667, "sat/validation": 0.7027027027027027, "u2/validation": 0.5833333333333334, "u4/validation": 0.5416666666666666, "google/validation": 1.0, "bats/validation": 0.8140703517587939, "t_rex_relational_similarity/validation": 0.3084677419354839, "conceptnet_relational_similarity/validation": 0.3300359712230216, "nell_relational_similarity/validation": 0.6125}
|
analogy.reverse.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"sat_full/test": 0.6256684491978609, "sat/test": 0.6409495548961425, "u2/test": 0.6535087719298246, "u4/test": 0.6620370370370371, "google/test": 0.96, "bats/test": 0.8148971650917176, "t_rex_relational_similarity/test": 0.6284153005464481, "conceptnet_relational_similarity/test": 0.3859060402684564, "nell_relational_similarity/test": 0.71, "sat/validation": 0.4864864864864865, "u2/validation": 0.625, "u4/validation": 0.6875, "google/validation": 1.0, "bats/validation": 0.8542713567839196, "semeval2012_relational_similarity/validation": 0.5949367088607594, "t_rex_relational_similarity/validation": 0.2963709677419355, "conceptnet_relational_similarity/validation": 0.31744604316546765, "nell_relational_similarity/validation": 0.655}
|
classification.json
ADDED
@@ -0,0 +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": 0.9171312339912611, "test/f1_macro": 0.9144771714929822, "test/f1_micro": 0.9171312339912611, "test/p_macro": 0.9140990494403654, "test/p_micro": 0.9171312339912611, "test/r_macro": 0.9153113450496887, "test/r_micro": 0.9171312339912611}, "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": 0.8455399061032863, "test/f1_macro": 0.6723727069325071, "test/f1_micro": 0.8455399061032864, "test/p_macro": 0.6945535744714355, "test/p_micro": 0.8455399061032863, "test/r_macro": 0.6549786745504992, "test/r_micro": 0.8455399061032863}, "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": 0.6793066088840737, "test/f1_macro": 0.6676508193568021, "test/f1_micro": 0.6793066088840737, "test/p_macro": 0.6791394688186779, "test/p_micro": 0.6793066088840737, "test/r_macro": 0.6641670647971617, "test/r_micro": 0.6793066088840737}, "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": 0.95875356472143, "test/f1_macro": 0.8812996852310372, "test/f1_micro": 0.95875356472143, "test/p_macro": 0.9026237933224279, "test/p_micro": 0.95875356472143, "test/r_macro": 0.8634287514873775, "test/r_micro": 0.95875356472143}, "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": 0.9072391099968662, "test/f1_macro": 0.9041555105317078, "test/f1_micro": 0.9072391099968662, "test/p_macro": 0.9055653859573632, "test/p_micro": 0.9072391099968662, "test/r_macro": 0.9033586700399313, "test/r_micro": 0.9072391099968662}}
|
config.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "
|
3 |
"architectures": [
|
4 |
"RobertaModel"
|
5 |
],
|
|
|
1 |
{
|
2 |
+
"_name_or_path": "roberta-large",
|
3 |
"architectures": [
|
4 |
"RobertaModel"
|
5 |
],
|
finetuning_config.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"template": "I wasn\u2019t aware of this relationship, but I just read in the encyclopedia that <subj> is the <mask> of <obj>",
|
3 |
+
"model": "roberta-large",
|
4 |
+
"max_length": 64,
|
5 |
+
"epoch": 10,
|
6 |
+
"batch": 32,
|
7 |
+
"random_seed": 0,
|
8 |
+
"lr": 5e-06,
|
9 |
+
"lr_warmup": 10,
|
10 |
+
"aggregation_mode": "average_no_mask",
|
11 |
+
"data": "relbert/semeval2012_relational_similarity",
|
12 |
+
"data_name": null,
|
13 |
+
"exclude_relation": null,
|
14 |
+
"split": "train",
|
15 |
+
"split_valid": "validation",
|
16 |
+
"loss_function": "iloob",
|
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
|
24 |
+
}
|
relation_mapping.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer_config.json
CHANGED
@@ -6,7 +6,7 @@
|
|
6 |
"errors": "replace",
|
7 |
"mask_token": "<mask>",
|
8 |
"model_max_length": 512,
|
9 |
-
"name_or_path": "
|
10 |
"pad_token": "<pad>",
|
11 |
"sep_token": "</s>",
|
12 |
"special_tokens_map_file": null,
|
|
|
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,
|