User Tools

Site Tools


pipeline:window:cryolo:training

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
pipeline:window:cryolo:training [2019/09/18 10:14]
twagner
pipeline:window:cryolo:training [2020/03/16 15:21]
twagner
Line 6: Line 6:
  
 <note tip> <note tip>
 +**Use a different or multiple GPUs**
 +
 In the //"Optional arguments"// tab you can change the GPU that should be used by crYOLO. If you have multiple GPUs (e.g. nvidia-smi lists GPU 0 and GPU 1) you can also use both by setting the GPU argument to '0 1' In the //"Optional arguments"// tab you can change the GPU that should be used by crYOLO. If you have multiple GPUs (e.g. nvidia-smi lists GPU 0 and GPU 1) you can also use both by setting the GPU argument to '0 1'
 </note> </note>
Line 11: Line 13:
  
 In the GUI you have to fill in the mandatory fields: In the GUI you have to fill in the mandatory fields:
-{{ :pipeline:window:cryolo:cryolo_training.png?700 |}}+{{ :pipeline:window:cryolo:cryolo_training_202003.png?700 |}}
  
 The default number of warmup epochs((One epoch is a complete pass through the training data.)) is fine as long as you don't want to refine an existing model. During the warmup training epochs it will not try to estimate the size of your particle, which helps crYOLO to converge. The default number of warmup epochs((One epoch is a complete pass through the training data.)) is fine as long as you don't want to refine an existing model. During the warmup training epochs it will not try to estimate the size of your particle, which helps crYOLO to converge.
  
 <note tip> <note tip>
 +**When does crYOLO stop the training?**
 +
 When you start the training, it will stop when the "loss" metric on the validation data does not improve 10 times in a row. This is typically enough. In case you want to give the training more time to find the best model can increase the "not changed in a row" parameter to a higher value by setting the //early// argument in the //"Optional arguments"// to, for example, 15. When you start the training, it will stop when the "loss" metric on the validation data does not improve 10 times in a row. This is typically enough. In case you want to give the training more time to find the best model can increase the "not changed in a row" parameter to a higher value by setting the //early// argument in the //"Optional arguments"// to, for example, 15.
 </note> </note>
 +
 +The final model will be written to disk as specified in //saved_weights_name// in your configuration file.
  
 <html> <html>
Line 25: Line 31:
 </html> </html>
  
 +<html>
 +<div style="background-color: LightCyan ; padding: 10px; border: 1px solid Black;"> 
 +<b>Alternative: Train crYOLO using the command line</b>
 +</div>
 +</html>
  
-The final model will be written to disk as specified in //saved_weights_name// in your configuration file. 
  
-<hidden **Alternative:** Train crYOLO using the command line>+<hidden initialState="visible">
 Navigate to the folder with ''config_cryolo.json'' file, ''train_image'' folder, etc. Navigate to the folder with ''config_cryolo.json'' file, ''train_image'' folder, etc.
  
Line 34: Line 44:
  
 <code> <code>
-cryolo_train.py -c config.json -w 5 -g 0+cryolo_train.py -c config_cryolo.json -w 5 -g 0
 </code> </code>
  
 The final model file will be written to disk. The final model file will be written to disk.
  
- 
- 
-<code> 
-cryolo_train.py -c config.json -w 3 -g 0 -e 15 
-</code> 
- 
-to the training command. 
 </hidden> </hidden>
 +
 +<html>
 +<div style="background-color: LightCyan ; padding: 10px; border: 1px solid Black;"> 
 +<b> </b>
 +</div>
 +</html>
pipeline/window/cryolo/training.txt ยท Last modified: 2020/03/16 15:21 by twagner