{ "source_checkpoint": "/root/models/47b-ddt-refiners-gemma-shifted-logitnorm/ep2-step105450-step.pt", "global_step": 105450, "weights_sha256": "b3e09551b2842e6f21956f0de4e8d1c5ad846dbbe88f2a704554fd8495a30f6e", "num_tensors": 598, "num_parameters": 2487506290, "dropped_training_only_keys": [ "encoder.mta.mask_token", "encoder.mta.time_embedding.0.weight", "encoder.mta.time_embedding.0.bias", "encoder.mta.time_embedding.2.weight", "encoder.mta.time_embedding.2.bias", "encoder.mta.time_projection.1.weight", "encoder.mta.time_projection.1.bias", "encoder.mta.blocks.0.modulation", "encoder.mta.blocks.0.norm1.weight", "encoder.mta.blocks.0.norm1.bias", "encoder.mta.blocks.0.self_attn.q.weight", "encoder.mta.blocks.0.self_attn.q.bias", "encoder.mta.blocks.0.self_attn.k.weight", "encoder.mta.blocks.0.self_attn.k.bias", "encoder.mta.blocks.0.self_attn.v.weight", "encoder.mta.blocks.0.self_attn.v.bias", "encoder.mta.blocks.0.self_attn.o.weight", "encoder.mta.blocks.0.self_attn.o.bias", "encoder.mta.blocks.0.self_attn.norm_q.weight", "encoder.mta.blocks.0.self_attn.norm_k.weight", "encoder.mta.blocks.0.self_attn.gate_proj.weight", "encoder.mta.blocks.0.self_attn.gate_proj.bias", "encoder.mta.blocks.0.norm2.weight", "encoder.mta.blocks.0.norm2.bias", "encoder.mta.blocks.0.ffn.w13.weight", "encoder.mta.blocks.0.ffn.w2.weight", "encoder.mta.blocks.1.modulation", "encoder.mta.blocks.1.norm1.weight", "encoder.mta.blocks.1.norm1.bias", "encoder.mta.blocks.1.self_attn.q.weight", "encoder.mta.blocks.1.self_attn.q.bias", "encoder.mta.blocks.1.self_attn.k.weight", "encoder.mta.blocks.1.self_attn.k.bias", "encoder.mta.blocks.1.self_attn.v.weight", "encoder.mta.blocks.1.self_attn.v.bias", "encoder.mta.blocks.1.self_attn.o.weight", "encoder.mta.blocks.1.self_attn.o.bias", "encoder.mta.blocks.1.self_attn.norm_q.weight", "encoder.mta.blocks.1.self_attn.norm_k.weight", "encoder.mta.blocks.1.self_attn.gate_proj.weight", "encoder.mta.blocks.1.self_attn.gate_proj.bias", "encoder.mta.blocks.1.norm2.weight", "encoder.mta.blocks.1.norm2.bias", "encoder.mta.blocks.1.ffn.w13.weight", "encoder.mta.blocks.1.ffn.w2.weight", "encoder.mta.head.0.weight", "encoder.mta.head.0.bias", "encoder.mta.head.1.weight", "encoder.mta.head.1.bias", "encoder.mta.head.3.weight", "encoder.mta.head.3.bias" ], "renamed_keys": [ [ "encoder.patch_embedding.proj1.weight", "encoder.patch_embed.proj1.weight" ], [ "encoder.patch_embedding.proj2.weight", "encoder.patch_embed.proj2.weight" ], [ "encoder.patch_embedding.proj2.bias", "encoder.patch_embed.proj2.bias" ], [ "encoder.time_embedding.0.weight", "encoder.time_embed.0.weight" ], [ "encoder.time_embedding.0.bias", "encoder.time_embed.0.bias" ], [ "encoder.time_embedding.2.weight", "encoder.time_embed.2.weight" ], [ "encoder.time_embedding.2.bias", "encoder.time_embed.2.bias" ], [ "encoder.time_projection.1.weight", "encoder.time_proj.1.weight" ], [ "encoder.time_projection.1.bias", "encoder.time_proj.1.bias" ], [ "encoder.text_embedding.0.weight", "encoder.text_proj.0.weight" ], [ "encoder.text_embedding.2.weight", "encoder.text_proj.2.weight" ], [ "encoder.blocks.0.self_attn.q.weight", "encoder.blocks.0.attn.q.weight" ], [ "encoder.blocks.0.self_attn.q.bias", "encoder.blocks.0.attn.q.bias" ], [ "encoder.blocks.0.self_attn.k.weight", "encoder.blocks.0.attn.k.weight" ], [ "encoder.blocks.0.self_attn.k.bias", "encoder.blocks.0.attn.k.bias" ], [ "encoder.blocks.0.self_attn.v.weight", "encoder.blocks.0.attn.v.weight" ], [ "encoder.blocks.0.self_attn.v.bias", "encoder.blocks.0.attn.v.bias" ], [ "encoder.blocks.0.self_attn.o.weight", "encoder.blocks.0.attn.o.weight" ], [ "encoder.blocks.0.self_attn.o.bias", "encoder.blocks.0.attn.o.bias" ], [ "encoder.blocks.0.self_attn.norm_q.weight", "encoder.blocks.0.attn.norm_q.weight" ], [ "encoder.blocks.0.self_attn.norm_k.weight", "encoder.blocks.0.attn.norm_k.weight" ], [ "encoder.blocks.0.self_attn.gate_proj.weight", "encoder.blocks.0.attn.gate_proj.weight" ], [ "encoder.blocks.0.self_attn.gate_proj.bias", "encoder.blocks.0.attn.gate_proj.bias" ], [ "encoder.blocks.1.self_attn.q.weight", "encoder.blocks.1.attn.q.weight" ], [ "encoder.blocks.1.self_attn.q.bias", "encoder.blocks.1.attn.q.bias" ], [ "encoder.blocks.1.self_attn.k.weight", "encoder.blocks.1.attn.k.weight" ], [ "encoder.blocks.1.self_attn.k.bias", "encoder.blocks.1.attn.k.bias" ], [ "encoder.blocks.1.self_attn.v.weight", "encoder.blocks.1.attn.v.weight" ], [ "encoder.blocks.1.self_attn.v.bias", "encoder.blocks.1.attn.v.bias" ], [ "encoder.blocks.1.self_attn.o.weight", "encoder.blocks.1.attn.o.weight" ], [ "encoder.blocks.1.self_attn.o.bias", "encoder.blocks.1.attn.o.bias" ], [ "encoder.blocks.1.self_attn.norm_q.weight", "encoder.blocks.1.attn.norm_q.weight" ], [ "encoder.blocks.1.self_attn.norm_k.weight", "encoder.blocks.1.attn.norm_k.weight" ], [ "encoder.blocks.1.self_attn.gate_proj.weight", "encoder.blocks.1.attn.gate_proj.weight" ], [ "encoder.blocks.1.self_attn.gate_proj.bias", "encoder.blocks.1.attn.gate_proj.bias" ], [ "encoder.blocks.2.self_attn.q.weight", "encoder.blocks.2.attn.q.weight" ], [ "encoder.blocks.2.self_attn.q.bias", "encoder.blocks.2.attn.q.bias" ], [ "encoder.blocks.2.self_attn.k.weight", "encoder.blocks.2.attn.k.weight" ], [ "encoder.blocks.2.self_attn.k.bias", "encoder.blocks.2.attn.k.bias" ], [ "encoder.blocks.2.self_attn.v.weight", "encoder.blocks.2.attn.v.weight" ], [ "encoder.blocks.2.self_attn.v.bias", "encoder.blocks.2.attn.v.bias" ], [ "encoder.blocks.2.self_attn.o.weight", "encoder.blocks.2.attn.o.weight" ], [ "encoder.blocks.2.self_attn.o.bias", "encoder.blocks.2.attn.o.bias" ], [ "encoder.blocks.2.self_attn.norm_q.weight", "encoder.blocks.2.attn.norm_q.weight" ], [ "encoder.blocks.2.self_attn.norm_k.weight", "encoder.blocks.2.attn.norm_k.weight" ], [ "encoder.blocks.2.self_attn.gate_proj.weight", "encoder.blocks.2.attn.gate_proj.weight" ], [ "encoder.blocks.2.self_attn.gate_proj.bias", "encoder.blocks.2.attn.gate_proj.bias" ], [ "encoder.blocks.3.self_attn.q.weight", "encoder.blocks.3.attn.q.weight" ], [ "encoder.blocks.3.self_attn.q.bias", "encoder.blocks.3.attn.q.bias" ], [ "encoder.blocks.3.self_attn.k.weight", "encoder.blocks.3.attn.k.weight" ], [ "encoder.blocks.3.self_attn.k.bias", "encoder.blocks.3.attn.k.bias" ], [ "encoder.blocks.3.self_attn.v.weight", "encoder.blocks.3.attn.v.weight" ], [ "encoder.blocks.3.self_attn.v.bias", "encoder.blocks.3.attn.v.bias" ], [ "encoder.blocks.3.self_attn.o.weight", "encoder.blocks.3.attn.o.weight" ], [ "encoder.blocks.3.self_attn.o.bias", "encoder.blocks.3.attn.o.bias" ], [ "encoder.blocks.3.self_attn.norm_q.weight", "encoder.blocks.3.attn.norm_q.weight" ], [ "encoder.blocks.3.self_attn.norm_k.weight", "encoder.blocks.3.attn.norm_k.weight" ], [ "encoder.blocks.3.self_attn.gate_proj.weight", "encoder.blocks.3.attn.gate_proj.weight" ], [ "encoder.blocks.3.self_attn.gate_proj.bias", "encoder.blocks.3.attn.gate_proj.bias" ], [ "encoder.blocks.4.self_attn.q.weight", "encoder.blocks.4.attn.q.weight" ], [ "encoder.blocks.4.self_attn.q.bias", "encoder.blocks.4.attn.q.bias" ], [ "encoder.blocks.4.self_attn.k.weight", "encoder.blocks.4.attn.k.weight" ], [ "encoder.blocks.4.self_attn.k.bias", "encoder.blocks.4.attn.k.bias" ], [ "encoder.blocks.4.self_attn.v.weight", "encoder.blocks.4.attn.v.weight" ], [ "encoder.blocks.4.self_attn.v.bias", "encoder.blocks.4.attn.v.bias" ], [ "encoder.blocks.4.self_attn.o.weight", "encoder.blocks.4.attn.o.weight" ], [ "encoder.blocks.4.self_attn.o.bias", "encoder.blocks.4.attn.o.bias" ], [ "encoder.blocks.4.self_attn.norm_q.weight", "encoder.blocks.4.attn.norm_q.weight" ], [ "encoder.blocks.4.self_attn.norm_k.weight", "encoder.blocks.4.attn.norm_k.weight" ], [ "encoder.blocks.4.self_attn.gate_proj.weight", "encoder.blocks.4.attn.gate_proj.weight" ], [ "encoder.blocks.4.self_attn.gate_proj.bias", "encoder.blocks.4.attn.gate_proj.bias" ], [ "encoder.blocks.5.self_attn.q.weight", "encoder.blocks.5.attn.q.weight" ], [ "encoder.blocks.5.self_attn.q.bias", "encoder.blocks.5.attn.q.bias" ], [ "encoder.blocks.5.self_attn.k.weight", "encoder.blocks.5.attn.k.weight" ], [ "encoder.blocks.5.self_attn.k.bias", "encoder.blocks.5.attn.k.bias" ], [ "encoder.blocks.5.self_attn.v.weight", "encoder.blocks.5.attn.v.weight" ], [ "encoder.blocks.5.self_attn.v.bias", "encoder.blocks.5.attn.v.bias" ], [ "encoder.blocks.5.self_attn.o.weight", "encoder.blocks.5.attn.o.weight" ], [ "encoder.blocks.5.self_attn.o.bias", "encoder.blocks.5.attn.o.bias" ], [ "encoder.blocks.5.self_attn.norm_q.weight", "encoder.blocks.5.attn.norm_q.weight" ], [ "encoder.blocks.5.self_attn.norm_k.weight", "encoder.blocks.5.attn.norm_k.weight" ], [ "encoder.blocks.5.self_attn.gate_proj.weight", "encoder.blocks.5.attn.gate_proj.weight" ], [ "encoder.blocks.5.self_attn.gate_proj.bias", "encoder.blocks.5.attn.gate_proj.bias" ], [ "encoder.blocks.6.self_attn.q.weight", "encoder.blocks.6.attn.q.weight" ], [ "encoder.blocks.6.self_attn.q.bias", "encoder.blocks.6.attn.q.bias" ], [ "encoder.blocks.6.self_attn.k.weight", "encoder.blocks.6.attn.k.weight" ], [ "encoder.blocks.6.self_attn.k.bias", "encoder.blocks.6.attn.k.bias" ], [ "encoder.blocks.6.self_attn.v.weight", "encoder.blocks.6.attn.v.weight" ], [ "encoder.blocks.6.self_attn.v.bias", "encoder.blocks.6.attn.v.bias" ], [ "encoder.blocks.6.self_attn.o.weight", "encoder.blocks.6.attn.o.weight" ], [ "encoder.blocks.6.self_attn.o.bias", "encoder.blocks.6.attn.o.bias" ], [ "encoder.blocks.6.self_attn.norm_q.weight", "encoder.blocks.6.attn.norm_q.weight" ], [ "encoder.blocks.6.self_attn.norm_k.weight", "encoder.blocks.6.attn.norm_k.weight" ], [ "encoder.blocks.6.self_attn.gate_proj.weight", "encoder.blocks.6.attn.gate_proj.weight" ], [ "encoder.blocks.6.self_attn.gate_proj.bias", "encoder.blocks.6.attn.gate_proj.bias" ], [ "encoder.blocks.7.self_attn.q.weight", "encoder.blocks.7.attn.q.weight" ], [ "encoder.blocks.7.self_attn.q.bias", "encoder.blocks.7.attn.q.bias" ], [ "encoder.blocks.7.self_attn.k.weight", "encoder.blocks.7.attn.k.weight" ], [ "encoder.blocks.7.self_attn.k.bias", "encoder.blocks.7.attn.k.bias" ], [ "encoder.blocks.7.self_attn.v.weight", "encoder.blocks.7.attn.v.weight" ], [ "encoder.blocks.7.self_attn.v.bias", "encoder.blocks.7.attn.v.bias" ], [ "encoder.blocks.7.self_attn.o.weight", "encoder.blocks.7.attn.o.weight" ], [ "encoder.blocks.7.self_attn.o.bias", "encoder.blocks.7.attn.o.bias" ], [ "encoder.blocks.7.self_attn.norm_q.weight", "encoder.blocks.7.attn.norm_q.weight" ], [ "encoder.blocks.7.self_attn.norm_k.weight", "encoder.blocks.7.attn.norm_k.weight" ], [ "encoder.blocks.7.self_attn.gate_proj.weight", "encoder.blocks.7.attn.gate_proj.weight" ], [ "encoder.blocks.7.self_attn.gate_proj.bias", "encoder.blocks.7.attn.gate_proj.bias" ], [ "encoder.blocks.8.self_attn.q.weight", "encoder.blocks.8.attn.q.weight" ], [ "encoder.blocks.8.self_attn.q.bias", "encoder.blocks.8.attn.q.bias" ], [ "encoder.blocks.8.self_attn.k.weight", "encoder.blocks.8.attn.k.weight" ], [ "encoder.blocks.8.self_attn.k.bias", "encoder.blocks.8.attn.k.bias" ], [ "encoder.blocks.8.self_attn.v.weight", "encoder.blocks.8.attn.v.weight" ], [ "encoder.blocks.8.self_attn.v.bias", "encoder.blocks.8.attn.v.bias" ], [ "encoder.blocks.8.self_attn.o.weight", "encoder.blocks.8.attn.o.weight" ], [ "encoder.blocks.8.self_attn.o.bias", "encoder.blocks.8.attn.o.bias" ], [ "encoder.blocks.8.self_attn.norm_q.weight", "encoder.blocks.8.attn.norm_q.weight" ], [ "encoder.blocks.8.self_attn.norm_k.weight", "encoder.blocks.8.attn.norm_k.weight" ], [ "encoder.blocks.8.self_attn.gate_proj.weight", "encoder.blocks.8.attn.gate_proj.weight" ], [ "encoder.blocks.8.self_attn.gate_proj.bias", "encoder.blocks.8.attn.gate_proj.bias" ], [ "encoder.blocks.9.self_attn.q.weight", "encoder.blocks.9.attn.q.weight" ], [ "encoder.blocks.9.self_attn.q.bias", "encoder.blocks.9.attn.q.bias" ], [ "encoder.blocks.9.self_attn.k.weight", "encoder.blocks.9.attn.k.weight" ], [ "encoder.blocks.9.self_attn.k.bias", "encoder.blocks.9.attn.k.bias" ], [ "encoder.blocks.9.self_attn.v.weight", "encoder.blocks.9.attn.v.weight" ], [ "encoder.blocks.9.self_attn.v.bias", "encoder.blocks.9.attn.v.bias" ], [ "encoder.blocks.9.self_attn.o.weight", "encoder.blocks.9.attn.o.weight" ], [ "encoder.blocks.9.self_attn.o.bias", "encoder.blocks.9.attn.o.bias" ], [ "encoder.blocks.9.self_attn.norm_q.weight", "encoder.blocks.9.attn.norm_q.weight" ], [ "encoder.blocks.9.self_attn.norm_k.weight", "encoder.blocks.9.attn.norm_k.weight" ], [ "encoder.blocks.9.self_attn.gate_proj.weight", "encoder.blocks.9.attn.gate_proj.weight" ], [ "encoder.blocks.9.self_attn.gate_proj.bias", "encoder.blocks.9.attn.gate_proj.bias" ], [ "encoder.blocks.10.self_attn.q.weight", "encoder.blocks.10.attn.q.weight" ], [ "encoder.blocks.10.self_attn.q.bias", "encoder.blocks.10.attn.q.bias" ], [ "encoder.blocks.10.self_attn.k.weight", "encoder.blocks.10.attn.k.weight" ], [ "encoder.blocks.10.self_attn.k.bias", "encoder.blocks.10.attn.k.bias" ], [ "encoder.blocks.10.self_attn.v.weight", "encoder.blocks.10.attn.v.weight" ], [ "encoder.blocks.10.self_attn.v.bias", "encoder.blocks.10.attn.v.bias" ], [ "encoder.blocks.10.self_attn.o.weight", "encoder.blocks.10.attn.o.weight" ], [ "encoder.blocks.10.self_attn.o.bias", "encoder.blocks.10.attn.o.bias" ], [ "encoder.blocks.10.self_attn.norm_q.weight", "encoder.blocks.10.attn.norm_q.weight" ], [ "encoder.blocks.10.self_attn.norm_k.weight", "encoder.blocks.10.attn.norm_k.weight" ], [ "encoder.blocks.10.self_attn.gate_proj.weight", "encoder.blocks.10.attn.gate_proj.weight" ], [ "encoder.blocks.10.self_attn.gate_proj.bias", "encoder.blocks.10.attn.gate_proj.bias" ], [ "encoder.text_refiner_blocks.0.norm1.weight", "encoder.text_refiner.0.norm1.weight" ], [ "encoder.text_refiner_blocks.0.norm1.bias", "encoder.text_refiner.0.norm1.bias" ], [ "encoder.text_refiner_blocks.0.self_attn.q.weight", "encoder.text_refiner.0.attn.q.weight" ], [ "encoder.text_refiner_blocks.0.self_attn.q.bias", "encoder.text_refiner.0.attn.q.bias" ], [ "encoder.text_refiner_blocks.0.self_attn.k.weight", "encoder.text_refiner.0.attn.k.weight" ], [ "encoder.text_refiner_blocks.0.self_attn.k.bias", "encoder.text_refiner.0.attn.k.bias" ], [ "encoder.text_refiner_blocks.0.self_attn.v.weight", "encoder.text_refiner.0.attn.v.weight" ], [ "encoder.text_refiner_blocks.0.self_attn.v.bias", "encoder.text_refiner.0.attn.v.bias" ], [ "encoder.text_refiner_blocks.0.self_attn.o.weight", "encoder.text_refiner.0.attn.o.weight" ], [ "encoder.text_refiner_blocks.0.self_attn.o.bias", "encoder.text_refiner.0.attn.o.bias" ], [ "encoder.text_refiner_blocks.0.self_attn.norm_q.weight", "encoder.text_refiner.0.attn.norm_q.weight" ], [ "encoder.text_refiner_blocks.0.self_attn.norm_k.weight", "encoder.text_refiner.0.attn.norm_k.weight" ], [ "encoder.text_refiner_blocks.0.self_attn.gate_proj.weight", "encoder.text_refiner.0.attn.gate_proj.weight" ], [ "encoder.text_refiner_blocks.0.self_attn.gate_proj.bias", "encoder.text_refiner.0.attn.gate_proj.bias" ], [ "encoder.text_refiner_blocks.0.norm2.weight", "encoder.text_refiner.0.norm2.weight" ], [ "encoder.text_refiner_blocks.0.norm2.bias", "encoder.text_refiner.0.norm2.bias" ], [ "encoder.text_refiner_blocks.0.ffn.w13.weight", "encoder.text_refiner.0.ffn.w13.weight" ], [ "encoder.text_refiner_blocks.0.ffn.w2.weight", "encoder.text_refiner.0.ffn.w2.weight" ], [ "encoder.text_refiner_blocks.1.norm1.weight", "encoder.text_refiner.1.norm1.weight" ], [ "encoder.text_refiner_blocks.1.norm1.bias", "encoder.text_refiner.1.norm1.bias" ], [ "encoder.text_refiner_blocks.1.self_attn.q.weight", "encoder.text_refiner.1.attn.q.weight" ], [ "encoder.text_refiner_blocks.1.self_attn.q.bias", "encoder.text_refiner.1.attn.q.bias" ], [ "encoder.text_refiner_blocks.1.self_attn.k.weight", "encoder.text_refiner.1.attn.k.weight" ], [ "encoder.text_refiner_blocks.1.self_attn.k.bias", "encoder.text_refiner.1.attn.k.bias" ], [ "encoder.text_refiner_blocks.1.self_attn.v.weight", "encoder.text_refiner.1.attn.v.weight" ], [ "encoder.text_refiner_blocks.1.self_attn.v.bias", "encoder.text_refiner.1.attn.v.bias" ], [ "encoder.text_refiner_blocks.1.self_attn.o.weight", "encoder.text_refiner.1.attn.o.weight" ], [ "encoder.text_refiner_blocks.1.self_attn.o.bias", "encoder.text_refiner.1.attn.o.bias" ], [ "encoder.text_refiner_blocks.1.self_attn.norm_q.weight", "encoder.text_refiner.1.attn.norm_q.weight" ], [ "encoder.text_refiner_blocks.1.self_attn.norm_k.weight", "encoder.text_refiner.1.attn.norm_k.weight" ], [ "encoder.text_refiner_blocks.1.self_attn.gate_proj.weight", "encoder.text_refiner.1.attn.gate_proj.weight" ], [ "encoder.text_refiner_blocks.1.self_attn.gate_proj.bias", "encoder.text_refiner.1.attn.gate_proj.bias" ], [ "encoder.text_refiner_blocks.1.norm2.weight", "encoder.text_refiner.1.norm2.weight" ], [ "encoder.text_refiner_blocks.1.norm2.bias", "encoder.text_refiner.1.norm2.bias" ], [ "encoder.text_refiner_blocks.1.ffn.w13.weight", "encoder.text_refiner.1.ffn.w13.weight" ], [ "encoder.text_refiner_blocks.1.ffn.w2.weight", "encoder.text_refiner.1.ffn.w2.weight" ], [ "encoder.noise_refiner_blocks.0.modulation", "encoder.image_refiner.0.modulation" ], [ "encoder.noise_refiner_blocks.0.norm1.weight", "encoder.image_refiner.0.norm1.weight" ], [ "encoder.noise_refiner_blocks.0.norm1.bias", "encoder.image_refiner.0.norm1.bias" ], [ "encoder.noise_refiner_blocks.0.self_attn.q.weight", "encoder.image_refiner.0.attn.q.weight" ], [ "encoder.noise_refiner_blocks.0.self_attn.q.bias", "encoder.image_refiner.0.attn.q.bias" ], [ "encoder.noise_refiner_blocks.0.self_attn.k.weight", "encoder.image_refiner.0.attn.k.weight" ], [ "encoder.noise_refiner_blocks.0.self_attn.k.bias", "encoder.image_refiner.0.attn.k.bias" ], [ "encoder.noise_refiner_blocks.0.self_attn.v.weight", "encoder.image_refiner.0.attn.v.weight" ], [ "encoder.noise_refiner_blocks.0.self_attn.v.bias", "encoder.image_refiner.0.attn.v.bias" ], [ "encoder.noise_refiner_blocks.0.self_attn.o.weight", "encoder.image_refiner.0.attn.o.weight" ], [ "encoder.noise_refiner_blocks.0.self_attn.o.bias", "encoder.image_refiner.0.attn.o.bias" ], [ "encoder.noise_refiner_blocks.0.self_attn.norm_q.weight", "encoder.image_refiner.0.attn.norm_q.weight" ], [ "encoder.noise_refiner_blocks.0.self_attn.norm_k.weight", "encoder.image_refiner.0.attn.norm_k.weight" ], [ "encoder.noise_refiner_blocks.0.self_attn.gate_proj.weight", "encoder.image_refiner.0.attn.gate_proj.weight" ], [ "encoder.noise_refiner_blocks.0.self_attn.gate_proj.bias", "encoder.image_refiner.0.attn.gate_proj.bias" ], [ "encoder.noise_refiner_blocks.0.norm2.weight", "encoder.image_refiner.0.norm2.weight" ], [ "encoder.noise_refiner_blocks.0.norm2.bias", "encoder.image_refiner.0.norm2.bias" ], [ "encoder.noise_refiner_blocks.0.ffn.w13.weight", "encoder.image_refiner.0.ffn.w13.weight" ], [ "encoder.noise_refiner_blocks.0.ffn.w2.weight", "encoder.image_refiner.0.ffn.w2.weight" ], [ "encoder.noise_refiner_blocks.1.modulation", "encoder.image_refiner.1.modulation" ], [ "encoder.noise_refiner_blocks.1.norm1.weight", "encoder.image_refiner.1.norm1.weight" ], [ "encoder.noise_refiner_blocks.1.norm1.bias", "encoder.image_refiner.1.norm1.bias" ], [ "encoder.noise_refiner_blocks.1.self_attn.q.weight", "encoder.image_refiner.1.attn.q.weight" ], [ "encoder.noise_refiner_blocks.1.self_attn.q.bias", "encoder.image_refiner.1.attn.q.bias" ], [ "encoder.noise_refiner_blocks.1.self_attn.k.weight", "encoder.image_refiner.1.attn.k.weight" ], [ "encoder.noise_refiner_blocks.1.self_attn.k.bias", "encoder.image_refiner.1.attn.k.bias" ], [ "encoder.noise_refiner_blocks.1.self_attn.v.weight", "encoder.image_refiner.1.attn.v.weight" ], [ "encoder.noise_refiner_blocks.1.self_attn.v.bias", "encoder.image_refiner.1.attn.v.bias" ], [ "encoder.noise_refiner_blocks.1.self_attn.o.weight", "encoder.image_refiner.1.attn.o.weight" ], [ "encoder.noise_refiner_blocks.1.self_attn.o.bias", "encoder.image_refiner.1.attn.o.bias" ], [ "encoder.noise_refiner_blocks.1.self_attn.norm_q.weight", "encoder.image_refiner.1.attn.norm_q.weight" ], [ "encoder.noise_refiner_blocks.1.self_attn.norm_k.weight", "encoder.image_refiner.1.attn.norm_k.weight" ], [ "encoder.noise_refiner_blocks.1.self_attn.gate_proj.weight", "encoder.image_refiner.1.attn.gate_proj.weight" ], [ "encoder.noise_refiner_blocks.1.self_attn.gate_proj.bias", "encoder.image_refiner.1.attn.gate_proj.bias" ], [ "encoder.noise_refiner_blocks.1.norm2.weight", "encoder.image_refiner.1.norm2.weight" ], [ "encoder.noise_refiner_blocks.1.norm2.bias", "encoder.image_refiner.1.norm2.bias" ], [ "encoder.noise_refiner_blocks.1.ffn.w13.weight", "encoder.image_refiner.1.ffn.w13.weight" ], [ "encoder.noise_refiner_blocks.1.ffn.w2.weight", "encoder.image_refiner.1.ffn.w2.weight" ], [ "encoder.head.head.weight", "encoder.head.proj.weight" ], [ "encoder.head.head.bias", "encoder.head.proj.bias" ], [ "decoder.patch_embedding.proj1.weight", "decoder.patch_embed.proj1.weight" ], [ "decoder.patch_embedding.proj2.weight", "decoder.patch_embed.proj2.weight" ], [ "decoder.patch_embedding.proj2.bias", "decoder.patch_embed.proj2.bias" ], [ "decoder.time_embedding.0.weight", "decoder.time_embed.0.weight" ], [ "decoder.time_embedding.0.bias", "decoder.time_embed.0.bias" ], [ "decoder.time_embedding.2.weight", "decoder.time_embed.2.weight" ], [ "decoder.time_embedding.2.bias", "decoder.time_embed.2.bias" ], [ "decoder.time_projection.1.weight", "decoder.time_proj.1.weight" ], [ "decoder.time_projection.1.bias", "decoder.time_proj.1.bias" ], [ "decoder.text_embedding.0.weight", "decoder.text_proj.0.weight" ], [ "decoder.text_embedding.2.weight", "decoder.text_proj.2.weight" ], [ "decoder.blocks.0.self_attn.q.weight", "decoder.blocks.0.attn.q.weight" ], [ "decoder.blocks.0.self_attn.q.bias", "decoder.blocks.0.attn.q.bias" ], [ "decoder.blocks.0.self_attn.k.weight", "decoder.blocks.0.attn.k.weight" ], [ "decoder.blocks.0.self_attn.k.bias", "decoder.blocks.0.attn.k.bias" ], [ "decoder.blocks.0.self_attn.v.weight", "decoder.blocks.0.attn.v.weight" ], [ "decoder.blocks.0.self_attn.v.bias", "decoder.blocks.0.attn.v.bias" ], [ "decoder.blocks.0.self_attn.o.weight", "decoder.blocks.0.attn.o.weight" ], [ "decoder.blocks.0.self_attn.o.bias", "decoder.blocks.0.attn.o.bias" ], [ "decoder.blocks.0.self_attn.norm_q.weight", "decoder.blocks.0.attn.norm_q.weight" ], [ "decoder.blocks.0.self_attn.norm_k.weight", "decoder.blocks.0.attn.norm_k.weight" ], [ "decoder.blocks.0.self_attn.gate_proj.weight", "decoder.blocks.0.attn.gate_proj.weight" ], [ "decoder.blocks.0.self_attn.gate_proj.bias", "decoder.blocks.0.attn.gate_proj.bias" ], [ "decoder.blocks.1.self_attn.q.weight", "decoder.blocks.1.attn.q.weight" ], [ "decoder.blocks.1.self_attn.q.bias", "decoder.blocks.1.attn.q.bias" ], [ "decoder.blocks.1.self_attn.k.weight", "decoder.blocks.1.attn.k.weight" ], [ "decoder.blocks.1.self_attn.k.bias", "decoder.blocks.1.attn.k.bias" ], [ "decoder.blocks.1.self_attn.v.weight", "decoder.blocks.1.attn.v.weight" ], [ "decoder.blocks.1.self_attn.v.bias", "decoder.blocks.1.attn.v.bias" ], [ "decoder.blocks.1.self_attn.o.weight", "decoder.blocks.1.attn.o.weight" ], [ "decoder.blocks.1.self_attn.o.bias", "decoder.blocks.1.attn.o.bias" ], [ "decoder.blocks.1.self_attn.norm_q.weight", "decoder.blocks.1.attn.norm_q.weight" ], [ "decoder.blocks.1.self_attn.norm_k.weight", "decoder.blocks.1.attn.norm_k.weight" ], [ "decoder.blocks.1.self_attn.gate_proj.weight", "decoder.blocks.1.attn.gate_proj.weight" ], [ "decoder.blocks.1.self_attn.gate_proj.bias", "decoder.blocks.1.attn.gate_proj.bias" ], [ "decoder.blocks.2.self_attn.q.weight", "decoder.blocks.2.attn.q.weight" ], [ "decoder.blocks.2.self_attn.q.bias", "decoder.blocks.2.attn.q.bias" ], [ "decoder.blocks.2.self_attn.k.weight", "decoder.blocks.2.attn.k.weight" ], [ "decoder.blocks.2.self_attn.k.bias", "decoder.blocks.2.attn.k.bias" ], [ "decoder.blocks.2.self_attn.v.weight", "decoder.blocks.2.attn.v.weight" ], [ "decoder.blocks.2.self_attn.v.bias", "decoder.blocks.2.attn.v.bias" ], [ "decoder.blocks.2.self_attn.o.weight", "decoder.blocks.2.attn.o.weight" ], [ "decoder.blocks.2.self_attn.o.bias", "decoder.blocks.2.attn.o.bias" ], [ "decoder.blocks.2.self_attn.norm_q.weight", "decoder.blocks.2.attn.norm_q.weight" ], [ "decoder.blocks.2.self_attn.norm_k.weight", "decoder.blocks.2.attn.norm_k.weight" ], [ "decoder.blocks.2.self_attn.gate_proj.weight", "decoder.blocks.2.attn.gate_proj.weight" ], [ "decoder.blocks.2.self_attn.gate_proj.bias", "decoder.blocks.2.attn.gate_proj.bias" ], [ "decoder.blocks.3.self_attn.q.weight", "decoder.blocks.3.attn.q.weight" ], [ "decoder.blocks.3.self_attn.q.bias", "decoder.blocks.3.attn.q.bias" ], [ "decoder.blocks.3.self_attn.k.weight", "decoder.blocks.3.attn.k.weight" ], [ "decoder.blocks.3.self_attn.k.bias", "decoder.blocks.3.attn.k.bias" ], [ "decoder.blocks.3.self_attn.v.weight", "decoder.blocks.3.attn.v.weight" ], [ "decoder.blocks.3.self_attn.v.bias", "decoder.blocks.3.attn.v.bias" ], [ "decoder.blocks.3.self_attn.o.weight", "decoder.blocks.3.attn.o.weight" ], [ "decoder.blocks.3.self_attn.o.bias", "decoder.blocks.3.attn.o.bias" ], [ "decoder.blocks.3.self_attn.norm_q.weight", "decoder.blocks.3.attn.norm_q.weight" ], [ "decoder.blocks.3.self_attn.norm_k.weight", "decoder.blocks.3.attn.norm_k.weight" ], [ "decoder.blocks.3.self_attn.gate_proj.weight", "decoder.blocks.3.attn.gate_proj.weight" ], [ "decoder.blocks.3.self_attn.gate_proj.bias", "decoder.blocks.3.attn.gate_proj.bias" ], [ "decoder.blocks.4.self_attn.q.weight", "decoder.blocks.4.attn.q.weight" ], [ "decoder.blocks.4.self_attn.q.bias", "decoder.blocks.4.attn.q.bias" ], [ "decoder.blocks.4.self_attn.k.weight", "decoder.blocks.4.attn.k.weight" ], [ "decoder.blocks.4.self_attn.k.bias", "decoder.blocks.4.attn.k.bias" ], [ "decoder.blocks.4.self_attn.v.weight", "decoder.blocks.4.attn.v.weight" ], [ "decoder.blocks.4.self_attn.v.bias", "decoder.blocks.4.attn.v.bias" ], [ "decoder.blocks.4.self_attn.o.weight", "decoder.blocks.4.attn.o.weight" ], [ "decoder.blocks.4.self_attn.o.bias", "decoder.blocks.4.attn.o.bias" ], [ "decoder.blocks.4.self_attn.norm_q.weight", "decoder.blocks.4.attn.norm_q.weight" ], [ "decoder.blocks.4.self_attn.norm_k.weight", "decoder.blocks.4.attn.norm_k.weight" ], [ "decoder.blocks.4.self_attn.gate_proj.weight", "decoder.blocks.4.attn.gate_proj.weight" ], [ "decoder.blocks.4.self_attn.gate_proj.bias", "decoder.blocks.4.attn.gate_proj.bias" ], [ "decoder.blocks.5.self_attn.q.weight", "decoder.blocks.5.attn.q.weight" ], [ "decoder.blocks.5.self_attn.q.bias", "decoder.blocks.5.attn.q.bias" ], [ "decoder.blocks.5.self_attn.k.weight", "decoder.blocks.5.attn.k.weight" ], [ "decoder.blocks.5.self_attn.k.bias", "decoder.blocks.5.attn.k.bias" ], [ "decoder.blocks.5.self_attn.v.weight", "decoder.blocks.5.attn.v.weight" ], [ "decoder.blocks.5.self_attn.v.bias", "decoder.blocks.5.attn.v.bias" ], [ "decoder.blocks.5.self_attn.o.weight", "decoder.blocks.5.attn.o.weight" ], [ "decoder.blocks.5.self_attn.o.bias", "decoder.blocks.5.attn.o.bias" ], [ "decoder.blocks.5.self_attn.norm_q.weight", "decoder.blocks.5.attn.norm_q.weight" ], [ "decoder.blocks.5.self_attn.norm_k.weight", "decoder.blocks.5.attn.norm_k.weight" ], [ "decoder.blocks.5.self_attn.gate_proj.weight", "decoder.blocks.5.attn.gate_proj.weight" ], [ "decoder.blocks.5.self_attn.gate_proj.bias", "decoder.blocks.5.attn.gate_proj.bias" ], [ "decoder.blocks.6.self_attn.q.weight", "decoder.blocks.6.attn.q.weight" ], [ "decoder.blocks.6.self_attn.q.bias", "decoder.blocks.6.attn.q.bias" ], [ "decoder.blocks.6.self_attn.k.weight", "decoder.blocks.6.attn.k.weight" ], [ "decoder.blocks.6.self_attn.k.bias", "decoder.blocks.6.attn.k.bias" ], [ "decoder.blocks.6.self_attn.v.weight", "decoder.blocks.6.attn.v.weight" ], [ "decoder.blocks.6.self_attn.v.bias", "decoder.blocks.6.attn.v.bias" ], [ "decoder.blocks.6.self_attn.o.weight", "decoder.blocks.6.attn.o.weight" ], [ "decoder.blocks.6.self_attn.o.bias", "decoder.blocks.6.attn.o.bias" ], [ "decoder.blocks.6.self_attn.norm_q.weight", "decoder.blocks.6.attn.norm_q.weight" ], [ "decoder.blocks.6.self_attn.norm_k.weight", "decoder.blocks.6.attn.norm_k.weight" ], [ "decoder.blocks.6.self_attn.gate_proj.weight", "decoder.blocks.6.attn.gate_proj.weight" ], [ "decoder.blocks.6.self_attn.gate_proj.bias", "decoder.blocks.6.attn.gate_proj.bias" ], [ "decoder.blocks.7.self_attn.q.weight", "decoder.blocks.7.attn.q.weight" ], [ "decoder.blocks.7.self_attn.q.bias", "decoder.blocks.7.attn.q.bias" ], [ "decoder.blocks.7.self_attn.k.weight", "decoder.blocks.7.attn.k.weight" ], [ "decoder.blocks.7.self_attn.k.bias", "decoder.blocks.7.attn.k.bias" ], [ "decoder.blocks.7.self_attn.v.weight", "decoder.blocks.7.attn.v.weight" ], [ "decoder.blocks.7.self_attn.v.bias", "decoder.blocks.7.attn.v.bias" ], [ "decoder.blocks.7.self_attn.o.weight", "decoder.blocks.7.attn.o.weight" ], [ "decoder.blocks.7.self_attn.o.bias", "decoder.blocks.7.attn.o.bias" ], [ "decoder.blocks.7.self_attn.norm_q.weight", "decoder.blocks.7.attn.norm_q.weight" ], [ "decoder.blocks.7.self_attn.norm_k.weight", "decoder.blocks.7.attn.norm_k.weight" ], [ "decoder.blocks.7.self_attn.gate_proj.weight", "decoder.blocks.7.attn.gate_proj.weight" ], [ "decoder.blocks.7.self_attn.gate_proj.bias", "decoder.blocks.7.attn.gate_proj.bias" ], [ "decoder.blocks.8.self_attn.q.weight", "decoder.blocks.8.attn.q.weight" ], [ "decoder.blocks.8.self_attn.q.bias", "decoder.blocks.8.attn.q.bias" ], [ "decoder.blocks.8.self_attn.k.weight", "decoder.blocks.8.attn.k.weight" ], [ "decoder.blocks.8.self_attn.k.bias", "decoder.blocks.8.attn.k.bias" ], [ "decoder.blocks.8.self_attn.v.weight", "decoder.blocks.8.attn.v.weight" ], [ "decoder.blocks.8.self_attn.v.bias", "decoder.blocks.8.attn.v.bias" ], [ "decoder.blocks.8.self_attn.o.weight", "decoder.blocks.8.attn.o.weight" ], [ "decoder.blocks.8.self_attn.o.bias", "decoder.blocks.8.attn.o.bias" ], [ "decoder.blocks.8.self_attn.norm_q.weight", "decoder.blocks.8.attn.norm_q.weight" ], [ "decoder.blocks.8.self_attn.norm_k.weight", "decoder.blocks.8.attn.norm_k.weight" ], [ "decoder.blocks.8.self_attn.gate_proj.weight", "decoder.blocks.8.attn.gate_proj.weight" ], [ "decoder.blocks.8.self_attn.gate_proj.bias", "decoder.blocks.8.attn.gate_proj.bias" ], [ "decoder.blocks.9.self_attn.q.weight", "decoder.blocks.9.attn.q.weight" ], [ "decoder.blocks.9.self_attn.q.bias", "decoder.blocks.9.attn.q.bias" ], [ "decoder.blocks.9.self_attn.k.weight", "decoder.blocks.9.attn.k.weight" ], [ "decoder.blocks.9.self_attn.k.bias", "decoder.blocks.9.attn.k.bias" ], [ "decoder.blocks.9.self_attn.v.weight", "decoder.blocks.9.attn.v.weight" ], [ "decoder.blocks.9.self_attn.v.bias", "decoder.blocks.9.attn.v.bias" ], [ "decoder.blocks.9.self_attn.o.weight", "decoder.blocks.9.attn.o.weight" ], [ "decoder.blocks.9.self_attn.o.bias", "decoder.blocks.9.attn.o.bias" ], [ "decoder.blocks.9.self_attn.norm_q.weight", "decoder.blocks.9.attn.norm_q.weight" ], [ "decoder.blocks.9.self_attn.norm_k.weight", "decoder.blocks.9.attn.norm_k.weight" ], [ "decoder.blocks.9.self_attn.gate_proj.weight", "decoder.blocks.9.attn.gate_proj.weight" ], [ "decoder.blocks.9.self_attn.gate_proj.bias", "decoder.blocks.9.attn.gate_proj.bias" ], [ "decoder.blocks.10.self_attn.q.weight", "decoder.blocks.10.attn.q.weight" ], [ "decoder.blocks.10.self_attn.q.bias", "decoder.blocks.10.attn.q.bias" ], [ "decoder.blocks.10.self_attn.k.weight", "decoder.blocks.10.attn.k.weight" ], [ "decoder.blocks.10.self_attn.k.bias", "decoder.blocks.10.attn.k.bias" ], [ "decoder.blocks.10.self_attn.v.weight", "decoder.blocks.10.attn.v.weight" ], [ "decoder.blocks.10.self_attn.v.bias", "decoder.blocks.10.attn.v.bias" ], [ "decoder.blocks.10.self_attn.o.weight", "decoder.blocks.10.attn.o.weight" ], [ "decoder.blocks.10.self_attn.o.bias", "decoder.blocks.10.attn.o.bias" ], [ "decoder.blocks.10.self_attn.norm_q.weight", "decoder.blocks.10.attn.norm_q.weight" ], [ "decoder.blocks.10.self_attn.norm_k.weight", "decoder.blocks.10.attn.norm_k.weight" ], [ "decoder.blocks.10.self_attn.gate_proj.weight", "decoder.blocks.10.attn.gate_proj.weight" ], [ "decoder.blocks.10.self_attn.gate_proj.bias", "decoder.blocks.10.attn.gate_proj.bias" ], [ "decoder.text_refiner_blocks.0.norm1.weight", "decoder.text_refiner.0.norm1.weight" ], [ "decoder.text_refiner_blocks.0.norm1.bias", "decoder.text_refiner.0.norm1.bias" ], [ "decoder.text_refiner_blocks.0.self_attn.q.weight", "decoder.text_refiner.0.attn.q.weight" ], [ "decoder.text_refiner_blocks.0.self_attn.q.bias", "decoder.text_refiner.0.attn.q.bias" ], [ "decoder.text_refiner_blocks.0.self_attn.k.weight", "decoder.text_refiner.0.attn.k.weight" ], [ "decoder.text_refiner_blocks.0.self_attn.k.bias", "decoder.text_refiner.0.attn.k.bias" ], [ "decoder.text_refiner_blocks.0.self_attn.v.weight", "decoder.text_refiner.0.attn.v.weight" ], [ "decoder.text_refiner_blocks.0.self_attn.v.bias", "decoder.text_refiner.0.attn.v.bias" ], [ "decoder.text_refiner_blocks.0.self_attn.o.weight", "decoder.text_refiner.0.attn.o.weight" ], [ "decoder.text_refiner_blocks.0.self_attn.o.bias", "decoder.text_refiner.0.attn.o.bias" ], [ "decoder.text_refiner_blocks.0.self_attn.norm_q.weight", "decoder.text_refiner.0.attn.norm_q.weight" ], [ "decoder.text_refiner_blocks.0.self_attn.norm_k.weight", "decoder.text_refiner.0.attn.norm_k.weight" ], [ "decoder.text_refiner_blocks.0.self_attn.gate_proj.weight", "decoder.text_refiner.0.attn.gate_proj.weight" ], [ "decoder.text_refiner_blocks.0.self_attn.gate_proj.bias", "decoder.text_refiner.0.attn.gate_proj.bias" ], [ "decoder.text_refiner_blocks.0.norm2.weight", "decoder.text_refiner.0.norm2.weight" ], [ "decoder.text_refiner_blocks.0.norm2.bias", "decoder.text_refiner.0.norm2.bias" ], [ "decoder.text_refiner_blocks.0.ffn.w13.weight", "decoder.text_refiner.0.ffn.w13.weight" ], [ "decoder.text_refiner_blocks.0.ffn.w2.weight", "decoder.text_refiner.0.ffn.w2.weight" ], [ "decoder.text_refiner_blocks.1.norm1.weight", "decoder.text_refiner.1.norm1.weight" ], [ "decoder.text_refiner_blocks.1.norm1.bias", "decoder.text_refiner.1.norm1.bias" ], [ "decoder.text_refiner_blocks.1.self_attn.q.weight", "decoder.text_refiner.1.attn.q.weight" ], [ "decoder.text_refiner_blocks.1.self_attn.q.bias", "decoder.text_refiner.1.attn.q.bias" ], [ "decoder.text_refiner_blocks.1.self_attn.k.weight", "decoder.text_refiner.1.attn.k.weight" ], [ "decoder.text_refiner_blocks.1.self_attn.k.bias", "decoder.text_refiner.1.attn.k.bias" ], [ "decoder.text_refiner_blocks.1.self_attn.v.weight", "decoder.text_refiner.1.attn.v.weight" ], [ "decoder.text_refiner_blocks.1.self_attn.v.bias", "decoder.text_refiner.1.attn.v.bias" ], [ "decoder.text_refiner_blocks.1.self_attn.o.weight", "decoder.text_refiner.1.attn.o.weight" ], [ "decoder.text_refiner_blocks.1.self_attn.o.bias", "decoder.text_refiner.1.attn.o.bias" ], [ "decoder.text_refiner_blocks.1.self_attn.norm_q.weight", "decoder.text_refiner.1.attn.norm_q.weight" ], [ "decoder.text_refiner_blocks.1.self_attn.norm_k.weight", "decoder.text_refiner.1.attn.norm_k.weight" ], [ "decoder.text_refiner_blocks.1.self_attn.gate_proj.weight", "decoder.text_refiner.1.attn.gate_proj.weight" ], [ "decoder.text_refiner_blocks.1.self_attn.gate_proj.bias", "decoder.text_refiner.1.attn.gate_proj.bias" ], [ "decoder.text_refiner_blocks.1.norm2.weight", "decoder.text_refiner.1.norm2.weight" ], [ "decoder.text_refiner_blocks.1.norm2.bias", "decoder.text_refiner.1.norm2.bias" ], [ "decoder.text_refiner_blocks.1.ffn.w13.weight", "decoder.text_refiner.1.ffn.w13.weight" ], [ "decoder.text_refiner_blocks.1.ffn.w2.weight", "decoder.text_refiner.1.ffn.w2.weight" ], [ "decoder.noise_refiner_blocks.0.modulation", "decoder.image_refiner.0.modulation" ], [ "decoder.noise_refiner_blocks.0.norm1.weight", "decoder.image_refiner.0.norm1.weight" ], [ "decoder.noise_refiner_blocks.0.norm1.bias", "decoder.image_refiner.0.norm1.bias" ], [ "decoder.noise_refiner_blocks.0.self_attn.q.weight", "decoder.image_refiner.0.attn.q.weight" ], [ "decoder.noise_refiner_blocks.0.self_attn.q.bias", "decoder.image_refiner.0.attn.q.bias" ], [ "decoder.noise_refiner_blocks.0.self_attn.k.weight", "decoder.image_refiner.0.attn.k.weight" ], [ "decoder.noise_refiner_blocks.0.self_attn.k.bias", "decoder.image_refiner.0.attn.k.bias" ], [ "decoder.noise_refiner_blocks.0.self_attn.v.weight", "decoder.image_refiner.0.attn.v.weight" ], [ "decoder.noise_refiner_blocks.0.self_attn.v.bias", "decoder.image_refiner.0.attn.v.bias" ], [ "decoder.noise_refiner_blocks.0.self_attn.o.weight", "decoder.image_refiner.0.attn.o.weight" ], [ "decoder.noise_refiner_blocks.0.self_attn.o.bias", "decoder.image_refiner.0.attn.o.bias" ], [ "decoder.noise_refiner_blocks.0.self_attn.norm_q.weight", "decoder.image_refiner.0.attn.norm_q.weight" ], [ "decoder.noise_refiner_blocks.0.self_attn.norm_k.weight", "decoder.image_refiner.0.attn.norm_k.weight" ], [ "decoder.noise_refiner_blocks.0.self_attn.gate_proj.weight", "decoder.image_refiner.0.attn.gate_proj.weight" ], [ "decoder.noise_refiner_blocks.0.self_attn.gate_proj.bias", "decoder.image_refiner.0.attn.gate_proj.bias" ], [ "decoder.noise_refiner_blocks.0.norm2.weight", "decoder.image_refiner.0.norm2.weight" ], [ "decoder.noise_refiner_blocks.0.norm2.bias", "decoder.image_refiner.0.norm2.bias" ], [ "decoder.noise_refiner_blocks.0.ffn.w13.weight", "decoder.image_refiner.0.ffn.w13.weight" ], [ "decoder.noise_refiner_blocks.0.ffn.w2.weight", "decoder.image_refiner.0.ffn.w2.weight" ], [ "decoder.noise_refiner_blocks.1.modulation", "decoder.image_refiner.1.modulation" ], [ "decoder.noise_refiner_blocks.1.norm1.weight", "decoder.image_refiner.1.norm1.weight" ], [ "decoder.noise_refiner_blocks.1.norm1.bias", "decoder.image_refiner.1.norm1.bias" ], [ "decoder.noise_refiner_blocks.1.self_attn.q.weight", "decoder.image_refiner.1.attn.q.weight" ], [ "decoder.noise_refiner_blocks.1.self_attn.q.bias", "decoder.image_refiner.1.attn.q.bias" ], [ "decoder.noise_refiner_blocks.1.self_attn.k.weight", "decoder.image_refiner.1.attn.k.weight" ], [ "decoder.noise_refiner_blocks.1.self_attn.k.bias", "decoder.image_refiner.1.attn.k.bias" ], [ "decoder.noise_refiner_blocks.1.self_attn.v.weight", "decoder.image_refiner.1.attn.v.weight" ], [ "decoder.noise_refiner_blocks.1.self_attn.v.bias", "decoder.image_refiner.1.attn.v.bias" ], [ "decoder.noise_refiner_blocks.1.self_attn.o.weight", "decoder.image_refiner.1.attn.o.weight" ], [ "decoder.noise_refiner_blocks.1.self_attn.o.bias", "decoder.image_refiner.1.attn.o.bias" ], [ "decoder.noise_refiner_blocks.1.self_attn.norm_q.weight", "decoder.image_refiner.1.attn.norm_q.weight" ], [ "decoder.noise_refiner_blocks.1.self_attn.norm_k.weight", "decoder.image_refiner.1.attn.norm_k.weight" ], [ "decoder.noise_refiner_blocks.1.self_attn.gate_proj.weight", "decoder.image_refiner.1.attn.gate_proj.weight" ], [ "decoder.noise_refiner_blocks.1.self_attn.gate_proj.bias", "decoder.image_refiner.1.attn.gate_proj.bias" ], [ "decoder.noise_refiner_blocks.1.norm2.weight", "decoder.image_refiner.1.norm2.weight" ], [ "decoder.noise_refiner_blocks.1.norm2.bias", "decoder.image_refiner.1.norm2.bias" ], [ "decoder.noise_refiner_blocks.1.ffn.w13.weight", "decoder.image_refiner.1.ffn.w13.weight" ], [ "decoder.noise_refiner_blocks.1.ffn.w2.weight", "decoder.image_refiner.1.ffn.w2.weight" ], [ "decoder.head.head.weight", "decoder.head.proj.weight" ], [ "decoder.head.head.bias", "decoder.head.proj.bias" ] ] }