TastyPancakes
/
PhD-Notes
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

New experiment format

This commit is contained in:
Eduardo Cueto-Mendoza 2024-12-31 15:50:07 +00:00
parent a2c192d708
commit 6ee0fc76fb
Signed by: TastyPancakes
GPG Key ID: 941DF56C7242C3F1
1 changed files with 16 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
Experiment-1.norg
View File

@ -0,0 +1,16 @@
* Early Stopping initial experiment
With this experiment we try to verify empirically that as Early Stopping tolerance increases (to bigger values) we obtain the correct estimate of the correct epoch to stop training
- ( ) Train NN model at an increasing Early Stopping tolerance
- ( ) Repeat step one with bigger model sizes
- ( ) At every step capture the energy, performance, and entropy
- ( ) Record the stopping epoch
- ( ) Verify that there is a correlation between the stopping epochs and entropy
* Possible Results
We expect that the results show a correlation between an increase on complexity and stopping epoch i.e. as we increase Early Stopping tolenrance the complexity will either increase or decrease.
* Possible Conclusions
* Further Work