Comments (2)
Hi bạn,
với config phía dưới thì mình train được kết quả như log phía dưới bạn nhé
dataset_params = {
'name':'hw',
'data_root':'./data_line/',
'train_annotation':'train_line_annotation.txt',
'valid_annotation':'test_line_annotation.txt',
'image_height':64
}
params = {
'print_every':200,
'valid_every':15*200,
'iters':200000,
'checkpoint':'./checkpoint/transformerocr_checkpoint.pth',
'export':'./weights/transformerocr.pth',
'metrics': 10000,
'batch_size': 16
}
config['trainer'].update(params)
config['dataset'].update(dataset_params)
config['device'] = 'cuda:0'
iter: 195000 - valid loss: 0.759 - acc full seq: 0.3456 - acc per char: 0.8523
iter: 195200 - train loss: 0.674 - lr: 1.76e-06 - load time: 0.41 - gpu time: 221.51
iter: 195400 - train loss: 0.670 - lr: 1.61e-06 - load time: 0.03 - gpu time: 222.29
iter: 195600 - train loss: 0.677 - lr: 1.48e-06 - load time: 0.36 - gpu time: 221.27
iter: 195800 - train loss: 0.669 - lr: 1.35e-06 - load time: 0.36 - gpu time: 219.71
iter: 196000 - train loss: 0.673 - lr: 1.22e-06 - load time: 0.03 - gpu time: 222.98
iter: 196200 - train loss: 0.675 - lr: 1.10e-06 - load time: 0.36 - gpu time: 222.40
iter: 196400 - train loss: 0.668 - lr: 9.90e-07 - load time: 0.03 - gpu time: 219.35
iter: 196600 - train loss: 0.675 - lr: 8.84e-07 - load time: 0.36 - gpu time: 221.31
iter: 196800 - train loss: 0.669 - lr: 7.83e-07 - load time: 0.38 - gpu time: 222.39
iter: 197000 - train loss: 0.674 - lr: 6.89e-07 - load time: 0.04 - gpu time: 222.85
iter: 197200 - train loss: 0.677 - lr: 6.01e-07 - load time: 0.37 - gpu time: 221.09
iter: 197400 - train loss: 0.669 - lr: 5.18e-07 - load time: 0.41 - gpu time: 222.07
iter: 197600 - train loss: 0.675 - lr: 4.42e-07 - load time: 0.04 - gpu time: 221.97
iter: 197800 - train loss: 0.668 - lr: 3.72e-07 - load time: 0.38 - gpu time: 220.27
iter: 198000 - train loss: 0.675 - lr: 3.08e-07 - load time: 0.03 - gpu time: 221.95
iter: 198000 - valid loss: 0.761 - acc full seq: 0.3431 - acc per char: 0.8526
iter: 198200 - train loss: 0.674 - lr: 2.50e-07 - load time: 0.36 - gpu time: 220.71
iter: 198400 - train loss: 0.673 - lr: 1.99e-07 - load time: 0.37 - gpu time: 220.60
iter: 198600 - train loss: 0.679 - lr: 1.53e-07 - load time: 0.03 - gpu time: 220.39
iter: 198800 - train loss: 0.674 - lr: 1.13e-07 - load time: 0.38 - gpu time: 222.74
iter: 199000 - train loss: 0.670 - lr: 8.00e-08 - load time: 0.36 - gpu time: 220.68
iter: 199200 - train loss: 0.673 - lr: 5.26e-08 - load time: 0.03 - gpu time: 220.08
iter: 199400 - train loss: 0.673 - lr: 3.13e-08 - load time: 0.38 - gpu time: 219.46
iter: 199600 - train loss: 0.673 - lr: 1.61e-08 - load time: 0.03 - gpu time: 220.73
iter: 199800 - train loss: 0.678 - lr: 7.02e-09 - load time: 0.36 - gpu time: 220.84
iter: 200000 - train loss: 0.677 - lr: 4.00e-09 - load time: 0.38 - gpu time: 220.71
from vietocr.
@pbcquoc Cảm ơn bạn nhiều nhé. Một lần nữa cảm ơn bạn đã đóng góp vietocr cho cộng đồng :)
from vietocr.
Related Issues (20)
- IndexError: index out of range in self HOT 1
- Cannot pickle 'Environment' object HOT 4
- Extend Vocab HOT 2
- Error: train_hw: There is not enough space on the disk.
- Can we train VietOCR model for Japanese language ? HOT 2
- AttributeError: 'Delaunay' object has no attribute 'vertices' HOT 5
- lỗi load config HOT 1
- Có vấn đề khi sử dụng batch predict. HOT 1
- TypeError: 'Predictor' object is not callable
- Lỗi khi training custom dataset HOT 1
- Lỗi không chạy được sau khi cài HOT 4
- Predict trên 1 dataset lớn HOT 1
- Valid loss lớn hơn Training loss HOT 1
- Remove n invalid images HOT 2
- New PIL not AttributeError: module 'PIL.Image' has no attribute 'ANTIALIAS' HOT 6
- Load dataset HOT 6
- Augmentation Issues HOT 1
- Có thể giới hạn được các kí tự đầu ra không?
- infer with batch_size
- Hỏi cách huấn luyện trên tập dữ liệu tùy chỉnh
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from vietocr.