{ "metadata": { "total_size": 60272785536 }, "weight_map": { "blocks.0.attn.norm_k.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.attn.norm_q.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.attn.to_k.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.attn.to_out.bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.attn.to_out.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.attn.to_q.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.attn.to_v.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.ffn.experts.w1": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.ffn.experts.w2": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.ffn.experts.w3": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.ffn.router.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.norm1.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.norm2.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.norm_post_attn.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.norm_post_ffn.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.0.scale_shift_table": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.attn.norm_k.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.attn.norm_q.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.attn.to_k.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.attn.to_out.bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.attn.to_out.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.attn.to_q.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.attn.to_v.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.ffn.experts.w1": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.ffn.experts.w2": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.ffn.experts.w3": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.ffn.router.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.norm1.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.norm2.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.norm_post_attn.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.norm_post_ffn.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.1.scale_shift_table": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.10.attn.norm_k.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.attn.norm_q.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.attn.to_k.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.attn.to_out.bias": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.attn.to_out.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.attn.to_q.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.attn.to_v.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.ffn.experts.w1": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.ffn.experts.w2": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.ffn.experts.w3": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.ffn.router.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.norm1.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.norm2.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.norm_post_attn.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.norm_post_ffn.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.10.scale_shift_table": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.attn.norm_k.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.attn.norm_q.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.attn.to_k.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.attn.to_out.bias": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.attn.to_out.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.attn.to_q.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.attn.to_v.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.ffn.experts.w1": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.11.ffn.experts.w2": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.11.ffn.experts.w3": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.11.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.ffn.router.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.11.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.11.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.11.norm1.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.norm2.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.norm_post_attn.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.11.norm_post_ffn.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.11.scale_shift_table": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.12.attn.norm_k.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.attn.norm_q.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.attn.to_k.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.attn.to_out.bias": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.attn.to_out.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.attn.to_q.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.attn.to_v.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.ffn.experts.w1": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.ffn.experts.w2": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.ffn.experts.w3": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.ffn.router.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.norm1.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.norm2.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.norm_post_attn.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.norm_post_ffn.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.12.scale_shift_table": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.attn.norm_k.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.attn.norm_q.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.attn.to_k.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.attn.to_out.bias": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.attn.to_out.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.attn.to_q.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.attn.to_v.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.ffn.experts.w1": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.ffn.experts.w2": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.ffn.experts.w3": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.ffn.router.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.norm1.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.norm2.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.norm_post_attn.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.norm_post_ffn.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.13.scale_shift_table": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.attn.norm_k.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.attn.norm_q.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.attn.to_k.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.attn.to_out.bias": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.attn.to_out.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.attn.to_q.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.attn.to_v.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.ffn.experts.w1": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.ffn.experts.w2": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.ffn.experts.w3": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.ffn.router.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.norm1.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.norm2.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.norm_post_attn.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.norm_post_ffn.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.14.scale_shift_table": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.15.attn.norm_k.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.15.attn.norm_q.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.15.attn.to_k.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.15.attn.to_out.bias": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.attn.to_out.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.attn.to_q.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.15.attn.to_v.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.15.ffn.experts.w1": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.ffn.experts.w2": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.ffn.experts.w3": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.ffn.router.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.norm1.weight": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.15.norm2.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.norm_post_attn.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.norm_post_ffn.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.15.scale_shift_table": "diffusion_pytorch_model-00004-of-00013.safetensors", "blocks.16.attn.norm_k.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.attn.norm_q.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.attn.to_k.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.attn.to_out.bias": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.attn.to_out.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.attn.to_q.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.attn.to_v.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.ffn.experts.w1": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.ffn.experts.w2": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.ffn.experts.w3": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.ffn.router.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.norm1.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.norm2.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.norm_post_attn.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.norm_post_ffn.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.16.scale_shift_table": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.attn.norm_k.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.attn.norm_q.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.attn.to_k.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.attn.to_out.bias": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.attn.to_out.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.attn.to_q.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.attn.to_v.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.ffn.experts.w1": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.ffn.experts.w2": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.ffn.experts.w3": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.ffn.router.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.norm1.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.norm2.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.norm_post_attn.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.norm_post_ffn.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.17.scale_shift_table": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.attn.norm_k.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.attn.norm_q.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.attn.to_k.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.attn.to_out.bias": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.attn.to_out.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.attn.to_q.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.attn.to_v.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.ffn.experts.w1": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.ffn.experts.w2": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.ffn.experts.w3": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.ffn.router.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.norm1.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.norm2.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.norm_post_attn.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.norm_post_ffn.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.18.scale_shift_table": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.19.attn.norm_k.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.attn.norm_q.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.attn.to_k.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.19.attn.to_out.bias": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.attn.to_out.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.attn.to_q.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.19.attn.to_v.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.ffn.experts.w1": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.ffn.experts.w2": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.ffn.experts.w3": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.ffn.router.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.norm1.weight": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.19.norm2.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.norm_post_attn.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.norm_post_ffn.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.19.scale_shift_table": "diffusion_pytorch_model-00005-of-00013.safetensors", "blocks.2.attn.norm_k.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.attn.norm_q.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.attn.to_k.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.attn.to_out.bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.attn.to_out.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.attn.to_q.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.attn.to_v.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.ffn.experts.w1": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.ffn.experts.w2": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.ffn.experts.w3": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.ffn.router.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.norm1.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.norm2.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.norm_post_attn.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.norm_post_ffn.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.2.scale_shift_table": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.20.attn.norm_k.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.attn.norm_q.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.attn.to_k.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.attn.to_out.bias": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.attn.to_out.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.attn.to_q.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.attn.to_v.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.ffn.experts.w1": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.ffn.experts.w2": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.ffn.experts.w3": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.ffn.router.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.norm1.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.norm2.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.norm_post_attn.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.norm_post_ffn.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.20.scale_shift_table": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.attn.norm_k.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.attn.norm_q.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.attn.to_k.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.attn.to_out.bias": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.attn.to_out.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.attn.to_q.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.attn.to_v.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.ffn.experts.w1": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.ffn.experts.w2": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.ffn.experts.w3": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.ffn.router.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.norm1.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.norm2.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.norm_post_attn.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.norm_post_ffn.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.21.scale_shift_table": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.attn.norm_k.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.attn.norm_q.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.attn.to_k.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.attn.to_out.bias": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.attn.to_out.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.attn.to_q.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.attn.to_v.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.ffn.experts.w1": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.ffn.experts.w2": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.ffn.experts.w3": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.ffn.router.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.norm1.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.norm2.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.norm_post_attn.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.norm_post_ffn.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.22.scale_shift_table": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.23.attn.norm_k.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.attn.norm_q.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.attn.to_k.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.attn.to_out.bias": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.attn.to_out.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.attn.to_q.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.23.attn.to_v.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.ffn.experts.w1": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.ffn.experts.w2": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.ffn.experts.w3": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.ffn.router.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.norm1.weight": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.23.norm2.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.norm_post_attn.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.norm_post_ffn.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.23.scale_shift_table": "diffusion_pytorch_model-00006-of-00013.safetensors", "blocks.24.attn.norm_k.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.attn.norm_q.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.attn.to_k.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.attn.to_out.bias": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.attn.to_out.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.attn.to_q.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.attn.to_v.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.ffn.experts.w1": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.ffn.experts.w2": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.ffn.experts.w3": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.ffn.router.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.norm1.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.norm2.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.norm_post_attn.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.norm_post_ffn.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.24.scale_shift_table": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.attn.norm_k.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.attn.norm_q.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.attn.to_k.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.attn.to_out.bias": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.attn.to_out.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.attn.to_q.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.attn.to_v.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.ffn.experts.w1": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.ffn.experts.w2": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.ffn.experts.w3": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.ffn.router.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.norm1.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.norm2.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.norm_post_attn.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.norm_post_ffn.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.25.scale_shift_table": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.attn.norm_k.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.attn.norm_q.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.attn.to_k.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.attn.to_out.bias": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.attn.to_out.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.attn.to_q.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.attn.to_v.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.ffn.experts.w1": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.ffn.experts.w2": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.ffn.experts.w3": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.ffn.router.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.norm1.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.norm2.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.norm_post_attn.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.norm_post_ffn.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.26.scale_shift_table": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.27.attn.norm_k.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.attn.norm_q.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.attn.to_k.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.attn.to_out.bias": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.attn.to_out.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.attn.to_q.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.attn.to_v.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.ffn.experts.w1": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.ffn.experts.w2": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.ffn.experts.w3": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.ffn.router.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.norm1.weight": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.27.norm2.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.norm_post_attn.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.norm_post_ffn.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.27.scale_shift_table": "diffusion_pytorch_model-00007-of-00013.safetensors", "blocks.28.attn.norm_k.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.attn.norm_q.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.attn.to_k.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.attn.to_out.bias": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.attn.to_out.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.attn.to_q.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.attn.to_v.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.ffn.experts.w1": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.ffn.experts.w2": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.ffn.experts.w3": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.ffn.router.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.norm1.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.norm2.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.norm_post_attn.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.norm_post_ffn.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.28.scale_shift_table": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.attn.norm_k.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.attn.norm_q.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.attn.to_k.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.attn.to_out.bias": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.attn.to_out.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.attn.to_q.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.attn.to_v.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.ffn.experts.w1": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.ffn.experts.w2": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.ffn.experts.w3": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.ffn.router.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.norm1.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.norm2.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.norm_post_attn.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.norm_post_ffn.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.29.scale_shift_table": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.3.attn.norm_k.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.attn.norm_q.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.attn.to_k.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.attn.to_out.bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.attn.to_out.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.attn.to_q.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.attn.to_v.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.ffn.experts.w1": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.ffn.experts.w2": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.ffn.experts.w3": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.3.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.ffn.router.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.3.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.3.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.3.norm1.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.norm2.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.norm_post_attn.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.3.norm_post_ffn.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.3.scale_shift_table": "diffusion_pytorch_model-00001-of-00013.safetensors", "blocks.30.attn.norm_k.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.attn.norm_q.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.attn.to_k.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.attn.to_out.bias": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.attn.to_out.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.attn.to_q.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.attn.to_v.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.ffn.experts.w1": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.ffn.experts.w2": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.ffn.experts.w3": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.ffn.router.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.30.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.30.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.30.norm1.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.norm2.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.norm_post_attn.weight": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.30.norm_post_ffn.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.30.scale_shift_table": "diffusion_pytorch_model-00008-of-00013.safetensors", "blocks.31.attn.norm_k.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.attn.norm_q.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.attn.to_k.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.attn.to_out.bias": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.attn.to_out.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.attn.to_q.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.attn.to_v.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.ffn.experts.w1": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.ffn.experts.w2": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.ffn.experts.w3": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.ffn.router.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.norm1.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.norm2.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.norm_post_attn.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.norm_post_ffn.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.31.scale_shift_table": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.attn.norm_k.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.attn.norm_q.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.attn.to_k.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.attn.to_out.bias": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.attn.to_out.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.attn.to_q.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.attn.to_v.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.ffn.experts.w1": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.ffn.experts.w2": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.ffn.experts.w3": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.ffn.router.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.norm1.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.norm2.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.norm_post_attn.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.norm_post_ffn.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.32.scale_shift_table": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.attn.norm_k.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.attn.norm_q.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.attn.to_k.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.attn.to_out.bias": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.attn.to_out.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.attn.to_q.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.attn.to_v.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.ffn.experts.w1": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.ffn.experts.w2": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.ffn.experts.w3": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.ffn.router.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.norm1.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.norm2.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.norm_post_attn.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.norm_post_ffn.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.33.scale_shift_table": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.attn.norm_k.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.attn.norm_q.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.attn.to_k.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.attn.to_out.bias": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.attn.to_out.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.attn.to_q.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.attn.to_v.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.ffn.experts.w1": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.ffn.experts.w2": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.ffn.experts.w3": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.34.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.ffn.router.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.34.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.34.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.34.norm1.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.norm2.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.norm_post_attn.weight": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.34.norm_post_ffn.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.34.scale_shift_table": "diffusion_pytorch_model-00009-of-00013.safetensors", "blocks.35.attn.norm_k.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.attn.norm_q.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.attn.to_k.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.attn.to_out.bias": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.attn.to_out.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.attn.to_q.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.attn.to_v.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.ffn.experts.w1": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.ffn.experts.w2": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.ffn.experts.w3": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.ffn.router.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.norm1.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.norm2.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.norm_post_attn.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.norm_post_ffn.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.35.scale_shift_table": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.attn.norm_k.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.attn.norm_q.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.attn.to_k.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.attn.to_out.bias": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.attn.to_out.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.attn.to_q.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.attn.to_v.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.ffn.experts.w1": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.ffn.experts.w2": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.ffn.experts.w3": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.ffn.router.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.norm1.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.norm2.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.norm_post_attn.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.norm_post_ffn.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.36.scale_shift_table": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.attn.norm_k.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.attn.norm_q.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.attn.to_k.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.attn.to_out.bias": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.attn.to_out.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.attn.to_q.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.attn.to_v.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.ffn.experts.w1": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.ffn.experts.w2": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.ffn.experts.w3": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.ffn.router.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.norm1.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.norm2.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.norm_post_attn.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.norm_post_ffn.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.37.scale_shift_table": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.attn.norm_k.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.attn.norm_q.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.attn.to_k.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.attn.to_out.bias": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.attn.to_out.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.attn.to_q.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.attn.to_v.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.ffn.experts.w1": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.ffn.experts.w2": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.38.ffn.experts.w3": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.38.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.ffn.router.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.38.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.38.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.38.norm1.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.norm2.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.norm_post_attn.weight": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.38.norm_post_ffn.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.38.scale_shift_table": "diffusion_pytorch_model-00010-of-00013.safetensors", "blocks.39.attn.norm_k.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.attn.norm_q.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.attn.to_k.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.attn.to_out.bias": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.attn.to_out.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.attn.to_q.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.attn.to_v.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.ffn.experts.w1": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.ffn.experts.w2": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.ffn.experts.w3": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.ffn.router.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.norm1.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.norm2.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.norm_post_attn.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.norm_post_ffn.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.39.scale_shift_table": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.4.attn.norm_k.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.attn.norm_q.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.attn.to_k.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.attn.to_out.bias": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.attn.to_out.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.attn.to_q.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.attn.to_v.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.ffn.experts.w1": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.ffn.experts.w2": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.ffn.experts.w3": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.ffn.router.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.norm1.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.norm2.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.norm_post_attn.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.norm_post_ffn.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.4.scale_shift_table": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.40.attn.norm_k.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.attn.norm_q.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.attn.to_k.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.attn.to_out.bias": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.attn.to_out.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.attn.to_q.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.attn.to_v.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.ffn.experts.w1": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.ffn.experts.w2": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.ffn.experts.w3": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.ffn.router.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.norm1.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.norm2.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.norm_post_attn.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.norm_post_ffn.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.40.scale_shift_table": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.attn.norm_k.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.attn.norm_q.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.attn.to_k.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.attn.to_out.bias": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.attn.to_out.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.attn.to_q.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.attn.to_v.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.ffn.experts.w1": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.ffn.experts.w2": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.ffn.experts.w3": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.ffn.router.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.norm1.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.norm2.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.norm_post_attn.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.norm_post_ffn.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.41.scale_shift_table": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.attn.norm_k.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.attn.norm_q.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.attn.to_k.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.attn.to_out.bias": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.attn.to_out.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.attn.to_q.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.attn.to_v.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.ffn.experts.w1": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.42.ffn.experts.w2": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.42.ffn.experts.w3": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.42.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.ffn.router.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.42.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.42.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.42.norm1.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.norm2.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.norm_post_attn.weight": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.42.norm_post_ffn.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.42.scale_shift_table": "diffusion_pytorch_model-00011-of-00013.safetensors", "blocks.43.attn.norm_k.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.attn.norm_q.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.attn.to_k.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.attn.to_out.bias": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.attn.to_out.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.attn.to_q.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.attn.to_v.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.ffn.experts.w1": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.ffn.experts.w2": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.ffn.experts.w3": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.ffn.router.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.norm1.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.norm2.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.norm_post_attn.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.norm_post_ffn.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.43.scale_shift_table": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.attn.norm_k.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.attn.norm_q.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.attn.to_k.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.attn.to_out.bias": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.attn.to_out.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.attn.to_q.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.attn.to_v.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.ffn.experts.w1": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.ffn.experts.w2": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.ffn.experts.w3": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.ffn.router.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.norm1.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.norm2.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.norm_post_attn.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.norm_post_ffn.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.44.scale_shift_table": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.attn.norm_k.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.attn.norm_q.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.attn.to_k.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.attn.to_out.bias": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.attn.to_out.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.attn.to_q.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.attn.to_v.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.ffn.experts.w1": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.ffn.experts.w2": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.ffn.experts.w3": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.ffn.router.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.norm1.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.norm2.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.norm_post_attn.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.norm_post_ffn.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.45.scale_shift_table": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.46.attn.norm_k.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.46.attn.norm_q.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.46.attn.to_k.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.46.attn.to_out.bias": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.attn.to_out.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.attn.to_q.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.46.attn.to_v.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.46.ffn.experts.w1": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.ffn.experts.w2": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.ffn.experts.w3": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.ffn.router.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.norm1.weight": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.46.norm2.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.norm_post_attn.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.norm_post_ffn.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.46.scale_shift_table": "diffusion_pytorch_model-00012-of-00013.safetensors", "blocks.47.attn.norm_k.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.attn.norm_q.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.attn.to_k.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.attn.to_out.bias": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.attn.to_out.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.attn.to_q.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.attn.to_v.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.ffn.experts.w1": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.ffn.experts.w2": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.ffn.experts.w3": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.ffn.router.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.norm1.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.norm2.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.norm_post_attn.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.norm_post_ffn.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.47.scale_shift_table": "diffusion_pytorch_model-00013-of-00013.safetensors", "blocks.5.attn.norm_k.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.attn.norm_q.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.attn.to_k.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.attn.to_out.bias": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.attn.to_out.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.attn.to_q.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.attn.to_v.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.ffn.experts.w1": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.ffn.experts.w2": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.ffn.experts.w3": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.ffn.router.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.norm1.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.norm2.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.norm_post_attn.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.norm_post_ffn.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.5.scale_shift_table": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.attn.norm_k.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.attn.norm_q.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.attn.to_k.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.attn.to_out.bias": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.attn.to_out.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.attn.to_q.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.attn.to_v.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.ffn.experts.w1": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.ffn.experts.w2": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.ffn.experts.w3": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.ffn.router.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.norm1.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.norm2.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.norm_post_attn.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.norm_post_ffn.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.6.scale_shift_table": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.attn.norm_k.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.attn.norm_q.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.attn.to_k.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.attn.to_out.bias": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.attn.to_out.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.attn.to_q.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.attn.to_v.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.ffn.experts.w1": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.ffn.experts.w2": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.7.ffn.experts.w3": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.7.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.ffn.router.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.7.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.7.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.7.norm1.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.norm2.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.norm_post_attn.weight": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.7.norm_post_ffn.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.7.scale_shift_table": "diffusion_pytorch_model-00002-of-00013.safetensors", "blocks.8.attn.norm_k.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.attn.norm_q.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.attn.to_k.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.attn.to_out.bias": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.attn.to_out.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.attn.to_q.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.attn.to_v.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.ffn.experts.w1": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.ffn.experts.w2": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.ffn.experts.w3": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.ffn.router.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.norm1.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.norm2.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.norm_post_attn.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.norm_post_ffn.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.8.scale_shift_table": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.attn.norm_k.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.attn.norm_q.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.attn.to_k.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.attn.to_out.bias": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.attn.to_out.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.attn.to_q.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.attn.to_v.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.ffn.experts.w1": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.ffn.experts.w2": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.ffn.experts.w3": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.ffn.router.e_score_correction_bias": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.ffn.router.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.ffn.shared_experts.down_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.ffn.shared_experts.gate_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.ffn.shared_experts.up_proj.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.norm1.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.norm2.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.norm_post_attn.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.norm_post_ffn.weight": "diffusion_pytorch_model-00003-of-00013.safetensors", "blocks.9.scale_shift_table": "diffusion_pytorch_model-00003-of-00013.safetensors", "norm_out_modulation.1.bias": "diffusion_pytorch_model-00013-of-00013.safetensors", "norm_out_modulation.1.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "patch_embedder.bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "patch_embedder.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "proj_out.bias": "diffusion_pytorch_model-00013-of-00013.safetensors", "proj_out.weight": "diffusion_pytorch_model-00013-of-00013.safetensors", "text_embedder.linear_1.bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "text_embedder.linear_1.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "text_embedder.linear_2.bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "text_embedder.linear_2.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "text_embedder.norm.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "time_embedder.linear_1.bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "time_embedder.linear_1.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "time_embedder.linear_2.bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "time_embedder.linear_2.weight": "diffusion_pytorch_model-00001-of-00013.safetensors", "time_modulation.1.bias": "diffusion_pytorch_model-00001-of-00013.safetensors", "time_modulation.1.weight": "diffusion_pytorch_model-00001-of-00013.safetensors" } }