May i ask a question about CF #29
-
|
CF needs to be trained every time I use it, is it possible that I can scan directly with the results of the previous training mode after I have trained it once? |
Beta Was this translation helpful? Give feedback.
Replies: 3 comments
-
|
Under the condition of ensuring the correct rate, if i train the scan every time, the time is too long |
Beta Was this translation helpful? Give feedback.
-
|
Hi @DylanSang - actually, you shouldn't need to train CF but once (or even zero times if you use the models we supply). Once trained, you can use the model file it has generated each time for analysis. If you'd like, you can read our original MAPS '21 paper that describes the process here. In Figure 1, you perform Step 1 once (or never if you use our pre-built models). Each time you subsequently want to scan code, you perform Step 2. You can scan an entire directory, just select files, or even just a single file, depending on your needs. Does this make sense? Thanks for looking at ControlFlag! Best, |
Beta Was this translation helpful? Give feedback.
-
|
Can you make a example? when i use the cmd: scan_for_anomalies.sh , needs to a dir or a .ts file which use to -t; |
Beta Was this translation helpful? Give feedback.
Hi @DylanSang - actually, you shouldn't need to train CF but once (or even zero times if you use the models we supply). Once trained, you can use the model file it has generated each time for analysis.
If you'd like, you can read our original MAPS '21 paper that describes the process here. In Figure 1, you perform Step 1 once (or never if you use our pre-built models). Each time you subsequently want to scan code, you perform Step 2. You can scan an entire directory, just select files, or even just a single file, depending on your needs.
Does this make sense?
Thanks for looking at ControlFlag!
Best,
The ControlFlag Team