| { | |
| "_class_name": "AutoencoderKL", | |
| "_diffusers_version": "0.35.0", | |
| "_name_or_path": "pretrained_models/e2e-invae-hf", | |
| "act_fn": "silu", | |
| "block_out_channels": [ | |
| 128, | |
| 128, | |
| 256, | |
| 256, | |
| 512 | |
| ], | |
| "down_block_types": [ | |
| "DownEncoderBlock2D", | |
| "DownEncoderBlock2D", | |
| "DownEncoderBlock2D", | |
| "DownEncoderBlock2D", | |
| "AttnDownEncoderBlock2D" | |
| ], | |
| "force_upcast": true, | |
| "in_channels": 3, | |
| "latent_channels": 32, | |
| "latents_mean": [ | |
| -0.06789003312587738, | |
| -0.016108449548482895, | |
| 0.09119921177625656, | |
| -0.046135950833559036, | |
| 0.001249820226803422, | |
| -0.0013260077685117722, | |
| -0.07334873825311661, | |
| 0.02006705291569233, | |
| 0.037329867482185364, | |
| 0.05814030393958092, | |
| 0.027024362236261368, | |
| 0.05961156636476517, | |
| -0.06803550571203232, | |
| 0.000353501905919984, | |
| -0.009422995150089264, | |
| 0.061765942722558975, | |
| 0.07155892252922058, | |
| -0.018458519130945206, | |
| 0.05336713790893555, | |
| -0.03314107283949852, | |
| 0.01082497090101242, | |
| -6.244407995836809e-05, | |
| -0.07398278266191483, | |
| -0.0003959784808102995, | |
| 0.05076234042644501, | |
| -0.02650425210595131, | |
| 0.030632901936769485, | |
| -0.04680190980434418, | |
| 0.044710319489240646, | |
| 0.00409353245049715, | |
| -0.025997422635555267, | |
| -0.0413847342133522 | |
| ], | |
| "latents_std": [ | |
| 3.2336182594299316, | |
| 3.1916394233703613, | |
| 3.2679460048675537, | |
| 3.2960245609283447, | |
| 3.1971914768218994, | |
| 3.1818912029266357, | |
| 3.198270320892334, | |
| 3.239720106124878, | |
| 3.2053537368774414, | |
| 3.2395002841949463, | |
| 3.280738592147827, | |
| 3.210848808288574, | |
| 3.1949291229248047, | |
| 3.198901653289795, | |
| 3.192922592163086, | |
| 3.1897382736206055, | |
| 3.2126080989837646, | |
| 3.2152347564697266, | |
| 3.290510892868042, | |
| 3.192960262298584, | |
| 3.2884199619293213, | |
| 3.2654669284820557, | |
| 3.194502353668213, | |
| 3.287775754928589, | |
| 3.297902822494507, | |
| 3.2006747722625732, | |
| 3.2641217708587646, | |
| 3.210415840148926, | |
| 3.253077268600464, | |
| 3.1833975315093994, | |
| 3.205012321472168, | |
| 3.195270299911499 | |
| ], | |
| "layers_per_block": 2, | |
| "mid_block_add_attention": true, | |
| "norm_num_groups": 32, | |
| "out_channels": 3, | |
| "sample_size": 32, | |
| "scaling_factor": null, | |
| "shift_factor": null, | |
| "up_block_types": [ | |
| "AttnUpDecoderBlock2D", | |
| "UpDecoderBlock2D", | |
| "UpDecoderBlock2D", | |
| "UpDecoderBlock2D", | |
| "UpDecoderBlock2D" | |
| ], | |
| "use_post_quant_conv": true, | |
| "use_quant_conv": true | |
| } | |