how to reproduce you model used in the papar by training from scratch?

[Vandermode]

Hi, I dive into your code, and find the script provided (see below) seems doesn't match your noise flow model in the paper. Shall I disable the argument of --cam and --iso? Shall I add additional arguments e.g. --cond_gt and --init_sdn? Thanks~

python3 train_noise_flow.py --logdir noise_flow_model   --arch "sdn5|unc|unc|unc|unc|gain4|unc|unc|unc|unc" \
     --sidd_path './data/SIDD_Medium_Raw/Data' --cam IP --iso 800  --n_train_threads 16   \
     --width 4 --epochs 2000  --lr 1e-4 --n_batch_train 138 --n_batch_test 138 --epochs_full_valid 10 \
     --n_patches_per_image 2898 --patch_height 32 --patch_sampling uniform \
     --start_tr_im_idx 10   --end_tr_im_idx 12   --start_ts_im_idx 10   --end_ts_im_idx 12

[AbdoKamel]

To train the model in the paper, you can remove --cam and --iso as these are used only to specify a single camera or ISO level, respectively.

[SSRheart]

When using the released model for reproduce Marginal KL divergence between the generated and the real noise samples, i can only get 0.026( larger than 0.008 in paper).
I wonder if i made or omitted something mistakes?

[AbdoKamel]

When using the released model for reproduce Marginal KL divergence between the generated and the real noise samples, i can only get 0.026( larger than 0.008 in paper).
I wonder if i made or omitted something mistakes?

Hi, can you please specify which script did you use? Thanks!

[SSRheart]

When using the released model for reproduce Marginal KL divergence between the generated and the real noise samples, i can only get 0.026( larger than 0.008 in paper).
I wonder if i made or omitted something mistakes?

Hi, can you please specify which script did you use? Thanks!

Thanks for replying me~~
Here is the script i used. Continue_training means i restore the released model. I also commented out the training loop code and see the file 'sample.txt' for Marginal KL divergence.

python3 train_noise_flow.py --logdir noise_flow_model --arch "sdn5|unc|unc|unc|unc|gain4|unc|unc|unc|unc"
--sidd_path './data/SIDD_Medium_Raw/Data' --n_train_threads 16
--width 4 --epochs 2000 --lr 1e-4 --n_batch_train 138 --n_batch_test 138 --epochs_full_valid 10
--n_patches_per_image 2898 --patch_height 32 --patch_sampling uniform
--start_tr_im_idx 10 --end_tr_im_idx 12 --start_ts_im_idx 10 --end_ts_im_idx 12
--continue_training