Add model log parsers (GAN/RL/InstanceSeg/NLP)

[zzhfz]

Description

add unified log parsers for DCGAN, DQN, MaskRCNN and BERT

Usage Example

1.DCGAN

(venv_fasterrcnn) sunjinge@ares1:~/PyTorchModels/GAN/dcgan$ python dcgan_log_parser.py -l dcgan_train.log -o dcgan_metrics.csv
成功保存 5 条记录到 dcgan_metrics.csv
(venv_fasterrcnn) sunjinge@ares1:~/PyTorchModels/GAN/dcgan$ cat dcgan_metrics.csv
epoch,total_epochs,iteration,loss_d,loss_g,d_x,d_g_z
0,25,0,1.9049,4.0445,0.3691,0.4341
0,25,1,1.7582,5.5897,0.8462,0.7125
0,25,2,1.2323,6.233,0.6996,0.4707
0,25,3,1.088,4.9859,0.5757,0.1771
0,25,4,0.9804,6.2721,0.7919,0.4466

2.DQN

(venv_fasterrcnn) sunjinge@ares1:~/PyTorchModels/RL/dqn$ python dqn_log_parser.py -l dqn_train.log -o dqn_metrics.csv
成功保存 5 条训练记录到 dqn_metrics.csv
(venv_fasterrcnn) sunjinge@ares1:~/PyTorchModels/RL/dqn$ cat dqn_metrics.csv
timestamp,episode,loss,loss_smooth
2025-04-21 16:47:53,0,0.649788,0.649788
2025-04-21 16:47:57,10,0.000329,0.325058
2025-04-21 16:47:59,20,0.412298,0.354138
2025-04-21 16:48:02,30,0.817083,0.469874
2025-04-21 16:48:04,40,4.243031,1.224506

3.MaskRCNN

(venv_fasterrcnn) sunjinge@ares1:~/PyTorchModels/InstanceSeg/maskrcnn$ python maskrcnn_log_parser.py -l maskrcnn_train.log -o maskrcnn_metrics.csv
成功保存 7 条记录到 maskrcnn_metrics.csv
(venv_fasterrcnn) sunjinge@ares1:~/PyTorchModels/InstanceSeg/maskrcnn$ cat maskrcnn_metrics.csv
timestamp,type,iter,loss,loss_avg,mAP
2025-04-22 22:33:38,train,10,2.0061,2.5199,
,eval,,,,0.0
,eval,,,,0.0
,eval,,,,0.0
,eval,,,,0.0
,eval,,,,0.0

4.BERT

(venv_fasterrcnn) sunjinge@ares1:~/PyTorchModels/NLP/HuggingFace$ python bert_log_parser.py -l  bert_train.log -o bert_metrics.csv
成功保存 1 条记录到 bert_metrics.csv
(venv_fasterrcnn) sunjinge@ares1:~/PyTorchModels/NLP/HuggingFace$ cat bert_metrics.csv
exact,f1,total,HasAns_exact,HasAns_f1,HasAns_total,NoAns_exact,NoAns_f1,NoAns_total,best_exact,best_exact_thresh,best_f1,best_f1_thresh,epoch
50.07159100480081,50.07159100480081,11873,0.0,0.0,5928,100.0,100.0,5945,50.07159100480081,0.0,50.07159100480081,0.0,