Commit f9044b4b authored by hepj987's avatar hepj987
Browse files

补充精度数据

parent dbb44794
......@@ -192,7 +192,14 @@ sh bert_squad_gpus.sh
## 模型精度
待完善...
使用单张Z100收敛性如下所示:
| 模型任务 | 训练精度 |
| :----------------: | :------------------: |
| MNLI-class(单卡) | val_accuracy: 0.7387 |
| squad1.1(单卡) | F1-score:0.916378 |
# 源码仓库及问题反馈
......
......@@ -8,7 +8,7 @@ python3 run_classifier.py \
--init_checkpoint /public/home/hepj/model_source/bert-base-TF2/bert_model.ckpt \
--train_batch_size 320 \
--eval_batch_size 32 \
--steps_per_loop 1000 \
--steps_per_loop 10 \
--learning_rate 2e-5 \
--num_train_epochs 3 \
--num_gpus 1 \
......
......@@ -6,9 +6,9 @@ python3 run_classifier.py \
--eval_data_path /public/home/hepj/MNLI/eval.tf_record \
--bert_config_file /public/home/hepj/model_source/uncased_L-12_H-768_A-12/bert_config.json \
--init_checkpoint /public/home/hepj/model_source/bert-base-TF2/bert_model.ckpt \
--train_batch_size 320 \
--train_batch_size 1280 \
--eval_batch_size 32 \
--steps_per_loop 1000 \
--steps_per_loop 10 \
--learning_rate 2e-5 \
--num_train_epochs 3 \
--num_gpus 4 \
......
......@@ -7,7 +7,7 @@ python3 run_squad_xuan.py \
--train_data_path /public/home/hepj/model/tf2.7.0_Bert/squad1.1/train.tf_record \
--predict_file /public/home/hepj/model/model_source/sq1.1/dev-v1.1.json \
--init_checkpoint /public/home/hepj/model_source/bert-large-TF2/bert_model.ckpt \
--train_batch_size 4 \
--train_batch_size 16 \
--predict_batch_size 4 \
--learning_rate 2e-5 \
--log_steps 1 \
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment