README.md

metadata

datasets:
  - relbert/semeval2012_relational_similarity_v6
model-index:
  - name: relbert/relbert-roberta-base-semeval2012-v6-mask-prompt-a-nce-0
    results:
      - task:
          name: Relation Mapping
          type: sorting-task
        dataset:
          name: Relation Mapping
          args: relbert/relation_mapping
          type: relation-mapping
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.746468253968254
      - task:
          name: Analogy Questions (SAT full)
          type: multiple-choice-qa
        dataset:
          name: SAT full
          args: relbert/analogy_questions
          type: analogy-questions
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.5561497326203209
      - task:
          name: Analogy Questions (SAT)
          type: multiple-choice-qa
        dataset:
          name: SAT
          args: relbert/analogy_questions
          type: analogy-questions
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.5667655786350149
      - task:
          name: Analogy Questions (BATS)
          type: multiple-choice-qa
        dataset:
          name: BATS
          args: relbert/analogy_questions
          type: analogy-questions
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.7092829349638688
      - task:
          name: Analogy Questions (Google)
          type: multiple-choice-qa
        dataset:
          name: Google
          args: relbert/analogy_questions
          type: analogy-questions
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.888
      - task:
          name: Analogy Questions (U2)
          type: multiple-choice-qa
        dataset:
          name: U2
          args: relbert/analogy_questions
          type: analogy-questions
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.543859649122807
      - task:
          name: Analogy Questions (U4)
          type: multiple-choice-qa
        dataset:
          name: U4
          args: relbert/analogy_questions
          type: analogy-questions
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.5532407407407407
      - task:
          name: Lexical Relation Classification (BLESS)
          type: classification
        dataset:
          name: BLESS
          args: relbert/lexical_relation_classification
          type: relation-classification
        metrics:
          - name: F1
            type: f1
            value: 0.9186379388277837
          - name: F1 (macro)
            type: f1_macro
            value: 0.9155808335650759
      - task:
          name: Lexical Relation Classification (CogALexV)
          type: classification
        dataset:
          name: CogALexV
          args: relbert/lexical_relation_classification
          type: relation-classification
        metrics:
          - name: F1
            type: f1
            value: 0.846244131455399
          - name: F1 (macro)
            type: f1_macro
            value: 0.6697612594207677
      - task:
          name: Lexical Relation Classification (EVALution)
          type: classification
        dataset:
          name: BLESS
          args: relbert/lexical_relation_classification
          type: relation-classification
        metrics:
          - name: F1
            type: f1
            value: 0.6663055254604551
          - name: F1 (macro)
            type: f1_macro
            value: 0.6524226839660551
      - task:
          name: Lexical Relation Classification (K&H+N)
          type: classification
        dataset:
          name: K&H+N
          args: relbert/lexical_relation_classification
          type: relation-classification
        metrics:
          - name: F1
            type: f1
            value: 0.9513806774709606
          - name: F1 (macro)
            type: f1_macro
            value: 0.8678963774341718
      - task:
          name: Lexical Relation Classification (ROOT09)
          type: classification
        dataset:
          name: ROOT09
          args: relbert/lexical_relation_classification
          type: relation-classification
        metrics:
          - name: F1
            type: f1
            value: 0.8987778125979317
          - name: F1 (macro)
            type: f1_macro
            value: 0.8966892516191803

relbert/relbert-roberta-base-semeval2012-v6-mask-prompt-a-nce-0

RelBERT fine-tuned from roberta-base on
relbert/semeval2012_relational_similarity_v6. Fine-tuning is done via RelBERT library (see the repository for more detail). It achieves the following results on the relation understanding tasks:

• Analogy Question (dataset, full result):
  • Accuracy on SAT (full): 0.5561497326203209
  • Accuracy on SAT: 0.5667655786350149
  • Accuracy on BATS: 0.7092829349638688
  • Accuracy on U2: 0.543859649122807
  • Accuracy on U4: 0.5532407407407407
  • Accuracy on Google: 0.888
• Lexical Relation Classification (dataset, full result):
  • Micro F1 score on BLESS: 0.9186379388277837
  • Micro F1 score on CogALexV: 0.846244131455399
  • Micro F1 score on EVALution: 0.6663055254604551
  • Micro F1 score on K&H+N: 0.9513806774709606
  • Micro F1 score on ROOT09: 0.8987778125979317
• Relation Mapping (dataset, full result):
  • Accuracy on Relation Mapping: 0.746468253968254

Usage

This model can be used through the relbert library. Install the library via pip

pip install relbert

and activate model as below.

from relbert import RelBERT
model = RelBERT("relbert/relbert-roberta-base-semeval2012-v6-mask-prompt-a-nce-0")
vector = model.get_embedding(['Tokyo', 'Japan'])  # shape of (1024, )

Training hyperparameters

The following hyperparameters were used during training:

• model: "roberta-base"
• max_length: "64"
• mode: "mask"
• data: "relbert/semeval2012_relational_similarity_v6"
• split: "train"
• split_eval: "validation"
• template_mode: "manual"
• template: "Today, I finally discovered the relation between and : is the
• loss_function: "nce_logout"
• classification_loss: "False"
• temperature_nce_constant: "0.05"
• temperature_nce_rank: "{'min': 0.01, 'max': 0.05, 'type': 'linear'}"
• epoch: "8"
• batch: "128"
• lr: "5e-06"
• lr_decay: "False"
• lr_warmup: "1"
• weight_decay: "0"
• random_seed: "0"
• exclude_relation: "None"
• n_sample: "320"
• gradient_accumulation: "8"
• relation_level: "None"

The full configuration can be found at fine-tuning parameter file.

Reference

If you use any resource from RelBERT, please consider to cite our paper.

@inproceedings{ushio-etal-2021-distilling-relation-embeddings,
    title = "{D}istilling {R}elation {E}mbeddings from {P}re-trained {L}anguage {M}odels",
    author = "Ushio, Asahi  and
      Schockaert, Steven  and
      Camacho-Collados, Jose",
    booktitle = "EMNLP 2021",
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
}