Current Status

Note that this is an early release. Don't hesitate to report bugs/possible improvements! There are surely many.

Tibert

Tibert is a transformers-compatible reproduction from the paper End-to-end Neural Coreference Resolution with several modifications. Among these:

• Usage of BERT (or any BERT variant) as an encoder as in BERT for Coreference Resolution: Baselines and Analysis
• batch size can be greater than 1
• Support of singletons as in Adapted End-to-End Coreference Resolution System for Anaphoric Identities in Dialogues

It can be installed with pip install tibert.

Documentation

Simple Prediction Example

Here is an example of using the simple prediction interface:

from tibert import BertForCoreferenceResolution, predict_coref_simple
from tibert.utils import pprint_coreference_document
from transformers import BertTokenizerFast

model = BertForCoreferenceResolution.from_pretrained(
    "compnet-renard/bert-base-cased-literary-coref"
)
tokenizer = BertTokenizerFast.from_pretrained("bert-base-cased")

annotated_doc = predict_coref_simple(
    "Sli did not want the earpods. He didn't like them.", model, tokenizer
)

pprint_coreference_document(annotated_doc)

results in:

>>> (0 Sli ) did not want the earpods. (0 He ) didn't like them.

Batched Predictions for Performance

A more advanced prediction interface is available:

from transformers import BertTokenizerFast
from tibert import predict_coref, BertForCoreferenceResolution

model = BertForCoreferenceResolution.from_pretrained(
    "compnet-renard/bert-base-cased-literary-coref"
)
tokenizer = BertTokenizerFast.from_pretrained("bert-base-cased")

documents = [
    "Sli did not want the earpods. He didn't like them.",
    "Princess Liana felt sad, because Zarth Arn was gone. The princess went to sleep.",
]

annotated_docs = predict_coref(documents, model, tokenizer, batch_size=2)

for doc in annotated_docs:
    pprint_coreference_document(doc)

results in:

>>> (0 Sli ) did not want the earpods . (0 He ) didn't like them .

>>> (0 Princess Liana ) felt sad , because (1 Zarth Arn ) was gone . (0 The princess) went to sleep .

Training a model

Aside from the tibert.train.train_coref_model function, it is possible to train a model from the command line. Training a model requires installing the sacred library. Here is the most basic example:

python -m tibert.run_train with\
       dataset_path=/path/to/litbank/repository\
       out_model_path=/path/to/output/model/directory

The following parameters can be set (taken from ./tibert/run_train.py config function):

Parameter	Default Value
batch_size	1
epochs_nb	30
dataset_path	"~/litbank"
mentions_per_tokens	0.4
antecedents_nb	350
max_span_size	10
mention_scorer_hidden_size	3000
sents_per_documents_train	11
mention_loss_coeff	0.1
bert_lr	1e-5
task_lr	2e-4
dropout	0.3
segment_size	128
encoder	"bert-base-cased"
out_model_path	"~/tibert/model"

One can monitor training metrics by adding run observers using command line flags - see sacred documentation for more details.

Method

We reimplemented the model from Lee et al., 2017 from scratch, but used BERT as the encoder as in Joshi et al., 2019. We do not use higher order inference as in Lee et al., 2018 since it was found to be not necessarily useful by Xu and Choi, 2020.

Singletons

Unfortunately, the framework from Lee et al., 2017 cannot represent singletons. This is because the authors were working on the OntoNotes dataset, where singletons are not annotated. We wanted to work on Litbank, so we had to find a way to represent singletons.

We opted to do as in Xu and Choi, 2021: we consider mention with a high enough mention scores as singletons, even when they are in no clusters. To force the model to learn proper mention scores, we add an auxiliary loss on mention score (as in Xu and Choi, 2021). To counter dataset imbalance between positive and negative mentions, we opt to compute a weighted loss instead of performing sampling.

Additional Features

Several work make use of additional features. For now, only the distance between spans is implemented.

Results

The following table presents the results we obtained by training this model (for now, it has only one entry !). Note that:

• the reported result was obtained with a limitation were documents are truncated to 512 tokens, so they may not be accurate with the performance on full documents
• the reported results can not be directly compared to the performance in the original Litbank paper since we only compute performance on one split of the datas

Dataset	Base model	MUC	B3	CEAF	CoNLL F1
Litbank	bert-base-cased	75.49	65.69	55.56	65.58