Selected timesteps for ddim sampler: [ 1 21 41 61 81 101 121 141 161 181 201 221 241 261 281 301 321 341 361 381 401 421 441 461 481 501 521 541 561 581 601 621 641 661 681 701 721 741 761 781 801 821 841 861 881 901 921 941 961 981] Selected alphas for ddim sampler: a_t: tensor([0.9998, 0.9971, 0.9931, 0.9876, 0.9801, 0.9706, 0.9589, 0.9447, 0.9279, 0.9084, 0.8862, 0.8611, 0.8334, 0.8030, 0.7700, 0.7348, 0.6976, 0.6585, 0.6181, 0.5766, 0.5345, 0.4922, 0.4501, 0.4087, 0.3683, 0.3294, 0.2922, 0.2570, 0.2242, 0.1938, 0.1660, 0.1408, 0.1183, 0.0984, 0.0811, 0.0660, 0.0532, 0.0424, 0.0334, 0.0260, 0.0200, 0.0152, 0.0114, 0.0085, 0.0062, 0.0045, 0.0032, 0.0022, 0.0015, 0.0010]); a_(t-1): [0.99989998 0.99979734 0.99713886 0.99314541 0.98755956 0.98013663 0.97064948 0.9588933 0.94469088 0.92789799 0.90840852 0.88615978 0.86113644 0.83337462 0.80296397 0.77004915 0.73482919 0.6975556 0.65852839 0.61809063 0.57662094 0.53452456 0.49222353 0.45014533 0.40871155 0.3683264 0.32936576 0.29216716 0.2570214 0.22416589 0.19378017 0.16598386 0.14083675 0.11834133 0.09844732 0.08105777 0.06603664 0.05321705 0.04240997 0.03341274 0.02601707 0.02001623 0.01521103 0.01141463 0.00845604 0.00618227 0.00445942 0.00317273 0.00222579 0.00153924] For the chosen value of eta, which is 0.0, this results in the following sigma_t schedule for ddim sampler tensor([0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.], dtype=torch.float64) Global seed set to 0 Creating invisible watermark encoder (see https://github.com/ShieldMnt/invisible-watermark)... Data shape for DDIM sampling is (1, 4, 256, 256), eta 0.0 Running DDIM Sampling with 50 timesteps DDIM Sampler: 0%| | 0/50 [00:00