JointBERT

(Unofficial) Pytorch implementation of JointBERT: BERT for Joint Intent Classification and Slot Filling

Model Architecture

• Predict intent and slot at the same time from one BERT model (=Joint model)
• total_loss = intent_loss + coef * slot_loss (Change coef with --slot_loss_coef option)
• If you want to use CRF layer, give --use_crf option

Dependencies

• python>=3.6
• torch==1.6.0
• transformers==3.0.2
• seqeval==0.0.12
• pytorch-crf==0.7.2

Dataset

• The number of labels are based on the train dataset.
• Add UNK for labels (For intent and slot labels which are only shown in dev and test dataset)
• Add PAD for slot label

Training & Evaluation

$ python3 main.py --task {task_name} \
                  --model_type {model_type} \
                  --model_dir {model_dir_name} \
                  --do_train --do_eval \
                  --use_crf

# For ATIS
$ python3 main.py --task atis \
                  --model_type bert \
                  --model_dir atis_model \
                  --do_train --do_eval
# For Snips
$ python3 main.py --task snips \
                  --model_type bert \
                  --model_dir snips_model \
                  --do_train --do_eval

Prediction

$ python3 predict.py --input_file {INPUT_FILE_PATH} --output_file {OUTPUT_FILE_PATH} --model_dir {SAVED_CKPT_PATH}

Results

• Run 5 ~ 10 epochs (Record the best result)
• Only test with uncased model
• ALBERT xxlarge sometimes can't converge well for slot prediction.

Updates

• 2019/12/03: Add DistilBert and RoBERTa result
• 2019/12/14: Add Albert (large v1) result
• 2019/12/22: Available to predict sentences
• 2019/12/26: Add Albert (xxlarge v1) result
• 2019/12/29: Add CRF option
• 2019/12/30: Available to check sentence-level semantic frame accuracy
• 2020/01/23: Only show the result related with uncased model
• 2020/04/03: Update with new prediction code

References

• Huggingface Transformers
• pytorch-crf