transformers/examples/question-answering

SQuAD

Based on the script run_qa.py.

Note: This script only works with models that have a fast tokenizer (backed by the 🤗 Tokenizers library) as it uses special features of those tokenizers. You can check if your favorite model has a fast tokenizer in this table, if it doesn't you can still use the old version of the script.

The old version of this script can be found here.

run_qa.py allows you to fine-tune any model from our hub (as long as its architecture as a ForQuestionAnswering version in the library) on the SQUAD dataset or another question-answering dataset of the datasets library or your own csv/jsonlines files as long as they are structured the same way as SQUAD. You might need to tweak the data processing inside the script if your data is structured differently.

Note that if your dataset contains samples with no possible answers (like SQUAD version 2), you need to pass along the flag --version_2_with_negative.

Fine-tuning BERT on SQuAD1.0

This example code fine-tunes BERT on the SQuAD1.0 dataset. It runs in 24 min (with BERT-base) or 68 min (with BERT-large) on a single tesla V100 16GB.

python run_qa.py \
  --model_name_or_path bert-base-uncased \
  --dataset_name squad \
  --do_train \
  --do_eval \
  --per_device_train_batch_size 12 \
  --learning_rate 3e-5 \
  --num_train_epochs 2 \
  --max_seq_length 384 \
  --doc_stride 128 \
  --output_dir /tmp/debug_squad/

Training with the previously defined hyper-parameters yields the following results:

f1 = 88.52
exact_match = 81.22

Distributed training

Here is an example using distributed training on 8 V100 GPUs and Bert Whole Word Masking uncased model to reach a F1 > 93 on SQuAD1.1:

python -m torch.distributed.launch --nproc_per_node=8 ./examples/question-answering/run_squad.py \
    --model_name_or_path bert-large-uncased-whole-word-masking \
    --dataset_name squad \
    --do_train \
    --do_eval \
    --learning_rate 3e-5 \
    --num_train_epochs 2 \
    --max_seq_length 384 \
    --doc_stride 128 \
    --output_dir ./examples/models/wwm_uncased_finetuned_squad/ \
    --per_device_eval_batch_size=3   \
    --per_device_train_batch_size=3   \

Training with the previously defined hyper-parameters yields the following results:

f1 = 93.15
exact_match = 86.91

This fine-tuned model is available as a checkpoint under the reference bert-large-uncased-whole-word-masking-finetuned-squad.

Fine-tuning XLNet with beam search on SQuAD

This example code fine-tunes XLNet on both SQuAD1.0 and SQuAD2.0 dataset.

Command for SQuAD1.0:

python run_qa_beam_search.py \
    --model_name_or_path xlnet-large-cased \
    --dataset_name squad \
    --do_train \
    --do_eval \
    --learning_rate 3e-5 \
    --num_train_epochs 2 \
    --max_seq_length 384 \
    --doc_stride 128 \
    --output_dir ./wwm_cased_finetuned_squad/ \
    --per_device_eval_batch_size=4  \
    --per_device_train_batch_size=4   \
    --save_steps 5000

Command for SQuAD2.0:

export SQUAD_DIR=/path/to/SQUAD

python run_qa_beam_search.py \
    --model_name_or_path xlnet-large-cased \
    --dataset_name squad_v2 \
    --do_train \
    --do_eval \
    --version_2_with_negative \
    --learning_rate 3e-5 \
    --num_train_epochs 4 \
    --max_seq_length 384 \
    --doc_stride 128 \
    --output_dir ./wwm_cased_finetuned_squad/ \
    --per_device_eval_batch_size=2  \
    --per_device_train_batch_size=2   \
    --save_steps 5000

Larger batch size may improve the performance while costing more memory.

Results for SQuAD1.0 with the previously defined hyper-parameters:

{
"exact": 85.45884578997162,
"f1": 92.5974600601065,
"total": 10570,
"HasAns_exact": 85.45884578997162,
"HasAns_f1": 92.59746006010651,
"HasAns_total": 10570
}

Results for SQuAD2.0 with the previously defined hyper-parameters:

{
"exact": 80.4177545691906,
"f1": 84.07154997729623,
"total": 11873,
"HasAns_exact": 76.73751686909581,
"HasAns_f1": 84.05558584352873,
"HasAns_total": 5928,
"NoAns_exact": 84.0874684608915,
"NoAns_f1": 84.0874684608915,
"NoAns_total": 5945
}

Fine-tuning BERT on SQuAD1.0 with relative position embeddings

The following examples show how to fine-tune BERT models with different relative position embeddings. The BERT model bert-base-uncased was pretrained with default absolute position embeddings. We provide the following pretrained models which were pre-trained on the same training data (BooksCorpus and English Wikipedia) as in the BERT model training, but with different relative position embeddings.

• zhiheng-huang/bert-base-uncased-embedding-relative-key, trained from scratch with relative embedding proposed by Shaw et al., Self-Attention with Relative Position Representations
• zhiheng-huang/bert-base-uncased-embedding-relative-key-query, trained from scratch with relative embedding method 4 in Huang et al. Improve Transformer Models with Better Relative Position Embeddings
• zhiheng-huang/bert-large-uncased-whole-word-masking-embedding-relative-key-query, fine-tuned from model bert-large-uncased-whole-word-masking with 3 additional epochs with relative embedding method 4 in Huang et al. Improve Transformer Models with Better Relative Position Embeddings

Base models fine-tuning

export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
python -m torch.distributed.launch --nproc_per_node=8 ./examples/question-answering/run_squad.py \
    --model_name_or_path zhiheng-huang/bert-base-uncased-embedding-relative-key-query \
    --dataset_name squad \
    --do_train \
    --do_eval \
    --learning_rate 3e-5 \
    --num_train_epochs 2 \
    --max_seq_length 512 \
    --doc_stride 128 \
    --output_dir relative_squad \
    --per_device_eval_batch_size=60 \
    --per_device_train_batch_size=6

Training with the above command leads to the following results. It boosts the BERT default from f1 score of 88.52 to 90.54.

'exact': 83.6802270577105, 'f1': 90.54772098174814

The change of max_seq_length from 512 to 384 in the above command leads to the f1 score of 90.34. Replacing the above model zhiheng-huang/bert-base-uncased-embedding-relative-key-query with zhiheng-huang/bert-base-uncased-embedding-relative-key leads to the f1 score of 89.51. The changing of 8 gpus to one gpu training leads to the f1 score of 90.71.

Large models fine-tuning

export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
python -m torch.distributed.launch --nproc_per_node=8 ./examples/question-answering/run_squad.py \
    --model_name_or_path zhiheng-huang/bert-large-uncased-whole-word-masking-embedding-relative-key-query \
    --dataset_name squad \
    --do_train \
    --do_eval \
    --learning_rate 3e-5 \
    --num_train_epochs 2 \
    --max_seq_length 512 \
    --doc_stride 128 \
    --output_dir relative_squad \
    --per_gpu_eval_batch_size=6 \
    --per_gpu_train_batch_size=2 \
    --gradient_accumulation_steps 3

Training with the above command leads to the f1 score of 93.52, which is slightly better than the f1 score of 93.15 for bert-large-uncased-whole-word-masking.

SQuAD with the Tensorflow Trainer

python run_tf_squad.py \
    --model_name_or_path bert-base-uncased \
    --output_dir model \
    --max_seq_length 384 \
    --num_train_epochs 2 \
    --per_gpu_train_batch_size 8 \
    --per_gpu_eval_batch_size 16 \
    --do_train \
    --logging_dir logs \    
    --logging_steps 10 \
    --learning_rate 3e-5 \
    --doc_stride 128    

For the moment evaluation is not available in the Tensorflow Trainer only the training.