sergioburdisso/dialog2flow-joint-bert-base

Dialog2Flow joint target model (BERT-base)

This is the original D2F$_{joint}$ model introduced in the paper "Dialog2Flow: Pre-training Soft-Contrastive Action-Driven Sentence Embeddings for Automatic Dialog Flow Extraction" published in the EMNLP 2024 main conference.

Implementation-wise, this is a sentence-transformers model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or search.

Usage (Sentence-Transformers)

Using this model becomes easy when you have sentence-transformers installed:

pip install -U sentence-transformers

Then you can use the model like this:

from sentence_transformers import SentenceTransformer
sentences = ["your phone please", "okay may i have your telephone number please"]

model = SentenceTransformer('sergioburdisso/dialog2flow-joint-bert-base')
embeddings = model.encode(sentences)
print(embeddings)

Usage (HuggingFace Transformers)

Without sentence-transformers, you can use the model like this: First, you pass your input through the transformer model, then you have to apply the right pooling-operation on-top of the contextualized word embeddings.

from transformers import AutoTokenizer, AutoModel
import torch


#Mean Pooling - Take attention mask into account for correct averaging
def mean_pooling(model_output, attention_mask):
    token_embeddings = model_output[0] #First element of model_output contains all token embeddings
    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)


# Sentences we want sentence embeddings for
sentences = ['your phone please', 'okay may i have your telephone number please']

# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('sergioburdisso/dialog2flow-joint-bert-base')
model = AutoModel.from_pretrained('sergioburdisso/dialog2flow-joint-bert-base')

# Tokenize sentences
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')

# Compute token embeddings
with torch.no_grad():
    model_output = model(**encoded_input)

# Perform pooling. In this case, mean pooling.
sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])

print("Sentence embeddings:")
print(sentence_embeddings)

Training

The model was trained with the parameters:

DataLoader:

torch.utils.data.dataloader.DataLoader of length 363506 with parameters:

{'batch_size': 64, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}

Loss:

spretrainer.losses.LabeledContrastiveLoss.LabeledContrastiveLoss

DataLoader:

torch.utils.data.dataloader.DataLoader of length 49478 with parameters:

{'batch_size': 64, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}

Loss:

spretrainer.losses.LabeledContrastiveLoss.LabeledContrastiveLoss

Parameters of the fit()-Method:

{
    "epochs": 15,
    "evaluation_steps": 164,
    "evaluator": [
        "spretrainer.evaluation.FewShotClassificationEvaluator.FewShotClassificationEvaluator"
    ],
    "max_grad_norm": 1,
    "optimizer_class": "<class 'torch.optim.adamw.AdamW'>",
    "optimizer_params": {
        "lr": 3e-06
    },
    "scheduler": "WarmupLinear",
    "warmup_steps": 100,
    "weight_decay": 0.01
}

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 64, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
)

Citation

@inproceedings{burdisso-etal-2024-dialog2flow,
    title = "Dialog2Flow: Pre-training Soft-Contrastive Action-Driven Sentence Embeddings for Automatic Dialog Flow Extraction",
    author = "Burdisso, Sergio  and
      Madikeri, Srikanth  and
      Motlicek, Petr",
    booktitle = "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing",
    month = nov,
    year = "2024",
    address = "Miami",
    publisher = "Association for Computational Linguistics",
}

License

Copyright (c) 2024 Idiap Research Institute. MIT License.