Skip to content

flobotics/func_sign_prob

Repository files navigation

func_sign_prob

Check out ubuntu-20-04-scripts/train_models

user => your username in the system

Hints for running on cloud instances (gcp or aws). If your ssh connection crashes, the script runs on, re-connect and tail -f nohup.out to watch what the script is doing. python3 -u prints all python-print functions without buffering, else you would see no output directly with tail -f nohup.out

nohup python3 -u xx.py & disown $!

Create directory /home/user/git mkdir -p /home/user/git

Switch into the new directory cd /home/user/git

Clone this repo git clone https://github.com/flobotics/func_sign_prob.git

Switch into func_sign_prob/ubuntu-20-04-scripts directory cd unc_sign_prob/ubuntu-20-04-scripts

Build pickle files from ubuntu packages. Pickle files contain disassembly (att and intel) for functions of binaries in ubuntu packages. And more info. --> ds-builder-raw.py

Older files. --> ds-builder.py and ds-builder-raw-host.py and ds-builder-2-Copy1.ipynb


After building ds-builder-raw.py go into ubuntu-20-04-scripts/build_tf_dataset/second directory.

First run "python3 tokenize_att_disassembly.py" .Perhaps you need to create the directory /tmp/testtars and copy all/some pickle files from ubuntu-20-04-pickles/ directory into /tmp/testtars, then run again.

Second run "python3 get_vocab_size_and_seq_length.py"

Third run "python3 build_return_type_dict.py"

Then switch into ubuntu-20-04-scripts/build_tf_model/second and run "python3 build_tf_tokenize_model.py"


-- third directory Next try with separating every number into its single numbers.So e.g. 0x546 gets 0 x 5 4 6 so model can learn every big number because they all consist of 0-9. So 0-9 is a single word.

For testing: Go to ubuntu-20-04-scripts/build_tf_datasets/third. Then create e.g. /tmp/test directory and copy some files from ubuntu-20-04-pickles/a* /tmp/test

Then run "python3 tokenize_att_disassembly.py -p=/tmp/test"

Then run "python3 get_vocab_size_and_seq_length.py"

Then run "python3 build_return_type_dict.py"

Then run "python3 build_tfrecord_files.py"

Then switch to ../../build_tf_model/third/ and

Then run "python3 build_tf_model.py"

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published