When I evaluate the diffusion model with "bash ./scripts/eval_diffusion.sh" by applying the checkpoint you provide, it does not show the results of the metric, which is mainly due to the fact that the code in "models/latent_diffusion_model_eval.py" does not include the calculation of the SEVIRSkillScore metric, when I borrow the metric code from "models/non_ar_model.py", and replace the "data_dict['gt'] = sample_predictions.mean(dim=1)" in models/latent_diffusion_model_eval.py with groudtruth image to calculate the metric, I cannot reproduce the results in the paper. Could you provide the code of the evaluation metric of the total diffusion model for me? And I do not understand why you write the code "data_dict['pred'] = sample_predictions.mean(dim=1), data_dict['gt'] = sample_predictions.mean(dim=1)" in models/latent_diffusion_model_eval.py. When I replace the data_dict['gt'] with the original picture, I find that the generated image totally different from the ground truth image .
When I evaluate the diffusion model with "bash ./scripts/eval_diffusion.sh" by applying the checkpoint you provide, it does not show the results of the metric, which is mainly due to the fact that the code in "models/latent_diffusion_model_eval.py" does not include the calculation of the SEVIRSkillScore metric, when I borrow the metric code from "models/non_ar_model.py", and replace the "data_dict['gt'] = sample_predictions.mean(dim=1)" in models/latent_diffusion_model_eval.py with groudtruth image to calculate the metric, I cannot reproduce the results in the paper. Could you provide the code of the evaluation metric of the total diffusion model for me? And I do not understand why you write the code "data_dict['pred'] = sample_predictions.mean(dim=1), data_dict['gt'] = sample_predictions.mean(dim=1)" in models/latent_diffusion_model_eval.py. When I replace the data_dict['gt'] with the original picture, I find that the generated image totally different from the ground truth image .