search

connorcoley / rexgen_direct

Template-free prediction of organic reaction outcomes
GNU General Public License v3.0
151 stars 69 forks source link

What is the cBond_detailed training dataset for WLDN model #18

Open YunpengCai opened 4 years ago

YunpengCai commented 4 years ago

Hi, I am trying to retrain the WLDN model with our own datasets, but I don't know how the training dataset (train.cBond_detailed) is made.

This dataset seems to be the part of the predicted result from the WLN model, but I cannot find any script in the repo to do this.

Could you please explain this part to me?

best regards Yunpeng

connorcoley commented 4 years ago

Hi Yunpeng,

The notes for usage can be seen here: https://github.com/connorcoley/rexgen_direct/blob/master/rexgen_direct/core_wln_global/notes.txt

Line 36 has the instructions for getting train.cbond_detailed: python nntest_direct.py --test ../data/train.txt.proc --hidden 300 --depth 3 --model model-300-3-direct --checkpoint ckpt-140000 --verbose 1 --detailed 1 > model-300-3-direct/train.cbond_detailed