search

CUMLSec / trex

MIT License
89 stars 13 forks source link

Missing field file in data-src: op_pos_emb #6

Open ice-tong opened 3 years ago

ice-tong commented 3 years ago

Run preprocess pre-train data script got an error log：

python command/pretrain/preprocess_pretrain_10k.py

...
No such file or directory: 'data-src/pretrain_10k/train.op_pos_emb'****
...
peikexin9 commented 3 years ago

That's my bad @ice-tong. I'll upload the updated pertaining files by this week.

Jerr-boop commented 3 years ago

hello，I want to known whether the question was to be solved? Please tell how to solve it ,Thank you very much!@ice-tong @peikexin9 peikexin9

ice-tong commented 3 years ago

hello，I want to known whether the question was to be solved? Please tell how to solve it ,Thank you very much!@ice-tong @peikexin9 peikexin9

This issue had solved by this commit： https://github.com/CUMLSec/trex/commit/0d8b738066b2cc65f38f678101e10f0564e1df93

Jerr-boop commented 3 years ago

Thank you very much for your answer. I also want to ask you a question. I just want to use this algorithm. Without training, what should I do? I look forward to your reply again.

ice-tong commented 3 years ago

Thank you very much for your answer. I also want to ask you a question. I just want to use this algorithm. Without training, what should I do? I look forward to your reply again.

I am sorry that I have no idea about that. It seems like this repo is still under development. Since this repo just a codebase of training now, you can not use this for binary function compare directly.

ice-tong commented 3 years ago

Thank you very much for your answer. I also want to ask you a question. I just want to use this algorithm. Without training, what should I do? I look forward to your reply again.

A related issue here:

https://github.com/CUMLSec/trex/issues/1

peikexin9 commented 3 years ago

Yancong @ice-tong, thanks for helping with the answer :-)

Hi @Jerr-boop, I will upload the finetuned model and scripts for obtaining function embedding and compare their similarity in few weeks. If you are interested, you can take a look at https://github.com/CUMLSec/trex/blob/main/fairseq/models/trex/hub_interface.py. I will load the finetuned model and leverage the predict function here: https://github.com/CUMLSec/trex/blob/3ca09e2a20d9cc8380d0ba05ab90b6ed2456edc7/fairseq/models/trex/hub_interface.py#L112

I will try to get this module ready soon and update you. Thanks again for your interest!