说明文档

该模型是基于 bert-base pruneofa 90% 稀疏在 Squadv1 数据集上进行迁移学习的版本。

  eval_exact_match = 80.2933
  eval_f1          = 87.6788
  eval_samples     =   10784

训练

使用 https://github.com/IntelLabs/Model-Compression-Research-Package.git 参见 pruneofa-transfer-learning.sh

评估

export CUDA_VISIBLE_DEVICES=0

OUTDIR=eval-bert-base-squadv1-pruneofa-90pc-bt
WORKDIR=transformers/examples/pytorch/question-answering
cd $WORKDIR

nohup python run_qa.py  \
    --model_name_or_path vuiseng9/bert-base-squadv1-pruneofa-90pc-bt  \
    --dataset_name squad  \
    --do_eval  \
    --per_device_eval_batch_size 128  \
    --max_seq_length 384  \
    --doc_stride 128  \
    --overwrite_output_dir \
    --output_dir $OUTDIR 2>&1 | tee $OUTDIR/run.log &

vuiseng9/bert-base-squadv1-pruneofa-90pc-bt

作者 vuiseng9

question-answering transformers

↓ 1 ♥ 0

创建时间: 2022-03-02 23:29:05+00:00

更新时间: 2022-01-18 19:13:21+00:00

在 Hugging Face 上查看

文件 (39)

.gitattributes

README.md

all_results.json

args.bin

bert-base-squadv1-pruneofa-90pc-bt.onnx ONNX

checkpoint-56750/config.json

checkpoint-56750/optimizer.pt

checkpoint-56750/pruneofa_lt_pytorch_model.bin

checkpoint-56750/pytorch_model.bin

checkpoint-56750/rng_state.pth

checkpoint-56750/scheduler.pt

checkpoint-56750/sparsified_pytorch_model.bin

checkpoint-56750/special_tokens_map.json

checkpoint-56750/tokenizer.json

checkpoint-56750/tokenizer_config.json

checkpoint-56750/trainer_state.json

checkpoint-56750/training_args.bin

checkpoint-56750/vocab.txt

config.json

eval_nbest_predictions.json

eval_predictions.json

eval_results.json

final_pytorch_model.bin

layer_wise_sparsity_global_rate_70.20.csv

layer_wise_sparsity_global_rate_70.20.md

linear_layer_sparsity_85M_params_90.00_sparsity.csv

linear_layer_sparsity_85M_params_90.00_sparsity.md

pruneofa-transfer-learning.sh

pruning_config.json

pytorch_model.bin

raw_pytorch_model.bin

run.log

special_tokens_map.json

tokenizer.json

tokenizer_config.json

train_results.json

trainer_state.json

training_args.bin

vocab.txt