Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

p2ch11 我尝试运行training这个文件,并对LunaModel 略做修改,两个结果有些不同,我不知道哪一个结果更好? #97

Open
smd-1886 opened this issue Sep 3, 2022 · 0 comments

Comments

@smd-1886
Copy link

smd-1886 commented Sep 3, 2022

我想知道下面的两个结果哪一个更好一些?
下面的结果,是我对LunaModel进行修改后的.

2022-09-03 09:43:48,765 INFO pid:12876 main:081:initModel Using CUDA; 1 devices.

2022-09-03 09:43:49,618 INFO pid:12876 main:140:main Starting LunaTrainingApp, Namespace(batch_size=32, comment='dwlpt', epochs=1, num_workers=8, tb_prefix='p2ch11')

2022-09-03 09:38:40,854 INFO pid:11603 dsets:191:init <dsets.LunaDataset object at 0x7f6486b69340>: 51244 training samples

2022-09-03 09:38:40,856 INFO pid:11603 dsets:191:init <dsets.LunaDataset object at 0x7f63966d2fd0>: 5694 validation samples

2022-09-03 09:38:40,856 INFO pid:11603 main:147:main Epoch 1 of 1, 1602/178 batches of size 32*1

2022-09-03 09:38:40,856 WARNING pid:11603 util.util:219:enumerateWithEstimate E1 Training ----/1602, starting

2022-09-03 09:38:45,845 INFO pid:11603 util.util:236:enumerateWithEstimate E1 Training 64/1602, done at 2022-09-03 09:40:06, 0:01:23

2022-09-03 09:38:55,958 INFO pid:11603 util.util:236:enumerateWithEstimate E1 Training 256/1602, done at 2022-09-03 09:40:06, 0:01:23

2022-09-03 09:39:36,904 INFO pid:11603 util.util:236:enumerateWithEstimate E1 Training 1024/1602, done at 2022-09-03 09:40:07, 0:01:24

2022-09-03 09:40:08,145 WARNING pid:11603 util.util:249:enumerateWithEstimate E1 Training ----/1602, done at 2022-09-03 09:40:08

2022-09-03 09:40:08,282 INFO pid:11603 main:261:logMetrics E1 LunaTrainingApp

2022-09-03 09:40:08,282 INFO pid:11603 main:291:logMetrics E1 trn nan loss, 36.1% correct,

2022-09-03 09:40:08,282 INFO pid:11603 main:300:logMetrics E1 trn_neg nan loss, 36.0% correct (18433 of 51135)

2022-09-03 09:40:08,282 INFO pid:11603 main:311:logMetrics E1 trn_pos nan loss, 66.1% correct (72 of 109)

2022-09-03 09:40:08,289 WARNING pid:11603 util.util:219:enumerateWithEstimate E1 Validation ----/178, starting

2022-09-03 09:40:09,335 INFO pid:11603 util.util:236:enumerateWithEstimate E1 Validation 64/178, done at 2022-09-03 09:40:10, 0:00:02

2022-09-03 09:40:11,296 WARNING pid:11603 util.util:249:enumerateWithEstimate E1 Validation ----/178, done at 2022-09-03 09:40:11

2022-09-03 09:40:11,540 INFO pid:11603 main:261:logMetrics E1 LunaTrainingApp

2022-09-03 09:40:11,541 INFO pid:11603 main:291:logMetrics E1 val nan loss, 0.2% correct,

2022-09-03 09:40:11,541 INFO pid:11603 main:300:logMetrics E1 val_neg nan loss, 0.0% correct (0 of 5681)

2022-09-03 09:40:11,541 INFO pid:11603 main:311:logMetrics E1 val_pos nan loss, 100.0% correct (13 of 13)

Process finished with exit code 0

下面的结果是原LunaModel没有修改过的

2022-09-03 09:43:48,765 INFO pid:12876 main:081:initModel Using CUDA; 1 devices.

2022-09-03 09:43:49,618 INFO pid:12876 main:140:main Starting LunaTrainingApp, Namespace(batch_size=32, comment='dwlpt', epochs=1, num_workers=8, tb_prefix='p2ch11')

2022-09-03 09:43:50,876 INFO pid:12876 dsets:191:init <dsets.LunaDataset object at 0x7f4852361340>: 51244 training samples

2022-09-03 09:43:50,878 INFO pid:12876 dsets:191:init <dsets.LunaDataset object at 0x7f4761eccfd0>: 5694 validation samples

2022-09-03 09:43:50,878 INFO pid:12876 main:147:main Epoch 1 of 1, 1602/178 batches of size 32*1

2022-09-03 09:43:50,879 WARNING pid:12876 util.util:219:enumerateWithEstimate E1 Training ----/1602, starting

2022-09-03 09:43:55,816 INFO pid:12876 util.util:236:enumerateWithEstimate E1 Training 64/1602, done at 2022-09-03 09:45:16, 0:01:24

2022-09-03 09:44:05,928 INFO pid:12876 util.util:236:enumerateWithEstimate E1 Training 256/1602, done at 2022-09-03 09:45:16, 0:01:23

2022-09-03 09:44:46,939 INFO pid:12876 util.util:236:enumerateWithEstimate E1 Training 1024/1602, done at 2022-09-03 09:45:17, 0:01:24

2022-09-03 09:45:18,085 WARNING pid:12876 util.util:249:enumerateWithEstimate E1 Training ----/1602, done at 2022-09-03 09:45:18

2022-09-03 09:45:18,223 INFO pid:12876 main:261:logMetrics E1 LunaTrainingApp

2022-09-03 09:45:18,224 INFO pid:12876 main:291:logMetrics E1 trn 0.1799 loss, 99.7% correct,

2022-09-03 09:45:18,224 INFO pid:12876 main:300:logMetrics E1 trn_neg 0.1162 loss, 99.9% correct (51085 of 51135)

2022-09-03 09:45:18,224 INFO pid:12876 main:311:logMetrics E1 trn_pos 30.0633 loss, 0.0% correct (0 of 109)

2022-09-03 09:45:18,228 WARNING pid:12876 util.util:219:enumerateWithEstimate E1 Validation ----/178, starting

2022-09-03 09:45:19,233 INFO pid:12876 util.util:236:enumerateWithEstimate E1 Validation 64/178, done at 2022-09-03 09:45:20, 0:00:02

2022-09-03 09:45:21,192 WARNING pid:12876 util.util:249:enumerateWithEstimate E1 Validation ----/178, done at 2022-09-03 09:45:21

2022-09-03 09:45:21,443 INFO pid:12876 main:261:logMetrics E1 LunaTrainingApp

2022-09-03 09:45:21,443 INFO pid:12876 main:291:logMetrics E1 val 0.0162 loss, 99.8% correct,

2022-09-03 09:45:21,443 INFO pid:12876 main:300:logMetrics E1 val_neg 0.0020 loss, 100.0% correct (5681 of 5681)

2022-09-03 09:45:21,443 INFO pid:12876 main:311:logMetrics E1 val_pos 6.2100 loss, 0.0% correct (0 of 13)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant