name,shape,rank,dtype,precision,numel,bits,size_bytes
img_in.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
img_in.weight,"[3072, 64]",2,torch.bfloat16,BF16,196608,16,393216
norm_out.linear.bias,[6144],1,torch.bfloat16,BF16,6144,16,12288
norm_out.linear.weight,"[6144, 3072]",2,torch.bfloat16,BF16,18874368,16,37748736
proj_out.bias,[64],1,torch.bfloat16,BF16,64,16,128
proj_out.weight,"[64, 3072]",2,torch.bfloat16,BF16,196608,16,393216
time_text_embed.timestep_embedder.linear_1.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
time_text_embed.timestep_embedder.linear_1.weight,"[3072, 256]",2,torch.bfloat16,BF16,786432,16,1572864
time_text_embed.timestep_embedder.linear_2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
time_text_embed.timestep_embedder.linear_2.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.0.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.0.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.0.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.0.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.0.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.0.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.0.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.0.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.0.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.0.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.0.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.0.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.0.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.0.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.0.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.0.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.0.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.0.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.0.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.0.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.0.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.0.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.0.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.0.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.0.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.0.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.0.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.0.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.0.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.0.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.0.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.0.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.1.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.1.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.1.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.1.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.1.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.1.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.1.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.1.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.1.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.1.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.1.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.1.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.1.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.1.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.1.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.1.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.1.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.1.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.1.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.1.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.1.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.1.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.1.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.1.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.1.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.1.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.1.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.1.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.1.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.1.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.1.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.1.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.10.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.10.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.10.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.10.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.10.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.10.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.10.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.10.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.10.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.10.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.10.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.10.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.10.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.10.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.10.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.10.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.10.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.10.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.10.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.10.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.10.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.10.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.10.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.10.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.10.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.10.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.10.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.10.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.10.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.10.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.10.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.10.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.11.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.11.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.11.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.11.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.11.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.11.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.11.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.11.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.11.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.11.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.11.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.11.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.11.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.11.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.11.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.11.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.11.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.11.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.11.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.11.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.11.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.11.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.11.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.11.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.11.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.11.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.11.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.11.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.11.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.11.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.11.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.11.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.12.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.12.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.12.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.12.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.12.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.12.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.12.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.12.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.12.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.12.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.12.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.12.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.12.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.12.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.12.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.12.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.12.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.12.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.12.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.12.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.12.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.12.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.12.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.12.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.12.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.12.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.12.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.12.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.12.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.12.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.12.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.12.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.13.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.13.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.13.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.13.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.13.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.13.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.13.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.13.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.13.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.13.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.13.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.13.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.13.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.13.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.13.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.13.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.13.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.13.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.13.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.13.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.13.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.13.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.13.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.13.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.13.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.13.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.13.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.13.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.13.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.13.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.13.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.13.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.14.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.14.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.14.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.14.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.14.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.14.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.14.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.14.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.14.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.14.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.14.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.14.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.14.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.14.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.14.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.14.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.14.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.14.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.14.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.14.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.14.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.14.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.14.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.14.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.14.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.14.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.14.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.14.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.14.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.14.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.14.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.14.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.15.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.15.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.15.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.15.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.15.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.15.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.15.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.15.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.15.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.15.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.15.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.15.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.15.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.15.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.15.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.15.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.15.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.15.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.15.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.15.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.15.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.15.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.15.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.15.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.15.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.15.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.15.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.15.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.15.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.15.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.15.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.15.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.16.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.16.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.16.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.16.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.16.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.16.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.16.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.16.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.16.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.16.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.16.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.16.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.16.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.16.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.16.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.16.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.16.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.16.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.16.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.16.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.16.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.16.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.16.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.16.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.16.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.16.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.16.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.16.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.16.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.16.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.16.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.16.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.17.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.17.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.17.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.17.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.17.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.17.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.17.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.17.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.17.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.17.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.17.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.17.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.17.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.17.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.17.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.17.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.17.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.17.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.17.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.17.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.17.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.17.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.17.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.17.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.17.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.17.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.17.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.17.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.17.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.17.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.17.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.17.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.18.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.18.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.18.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.18.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.18.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.18.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.18.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.18.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.18.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.18.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.18.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.18.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.18.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.18.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.18.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.18.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.18.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.18.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.18.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.18.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.18.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.18.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.18.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.18.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.18.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.18.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.18.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.18.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.18.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.18.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.18.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.18.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.19.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.19.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.19.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.19.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.19.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.19.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.19.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.19.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.19.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.19.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.19.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.19.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.19.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.19.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.19.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.19.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.19.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.19.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.19.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.19.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.19.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.19.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.19.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.19.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.19.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.19.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.19.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.19.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.19.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.19.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.19.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.19.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.2.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.2.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.2.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.2.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.2.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.2.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.2.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.2.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.2.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.2.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.2.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.2.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.2.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.2.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.2.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.2.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.2.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.2.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.2.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.2.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.2.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.2.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.2.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.2.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.2.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.2.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.2.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.2.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.2.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.2.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.2.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.2.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.20.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.20.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.20.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.20.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.20.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.20.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.20.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.20.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.20.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.20.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.20.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.20.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.20.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.20.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.20.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.20.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.20.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.20.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.20.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.20.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.20.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.20.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.20.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.20.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.20.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.20.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.20.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.20.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.20.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.20.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.20.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.20.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.21.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.21.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.21.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.21.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.21.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.21.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.21.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.21.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.21.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.21.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.21.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.21.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.21.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.21.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.21.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.21.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.21.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.21.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.21.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.21.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.21.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.21.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.21.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.21.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.21.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.21.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.21.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.21.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.21.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.21.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.21.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.21.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.22.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.22.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.22.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.22.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.22.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.22.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.22.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.22.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.22.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.22.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.22.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.22.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.22.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.22.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.22.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.22.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.22.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.22.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.22.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.22.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.22.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.22.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.22.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.22.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.22.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.22.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.22.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.22.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.22.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.22.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.22.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.22.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.23.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.23.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.23.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.23.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.23.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.23.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.23.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.23.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.23.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.23.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.23.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.23.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.23.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.23.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.23.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.23.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.23.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.23.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.23.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.23.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.23.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.23.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.23.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.23.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.23.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.23.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.23.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.23.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.23.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.23.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.23.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.23.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.24.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.24.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.24.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.24.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.24.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.24.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.24.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.24.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.24.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.24.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.24.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.24.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.24.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.24.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.24.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.24.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.24.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.24.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.24.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.24.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.24.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.24.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.24.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.24.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.24.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.24.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.24.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.24.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.24.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.24.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.24.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.24.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.25.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.25.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.25.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.25.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.25.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.25.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.25.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.25.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.25.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.25.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.25.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.25.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.25.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.25.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.25.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.25.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.25.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.25.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.25.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.25.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.25.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.25.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.25.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.25.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.25.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.25.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.25.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.25.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.25.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.25.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.25.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.25.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.26.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.26.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.26.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.26.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.26.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.26.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.26.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.26.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.26.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.26.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.26.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.26.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.26.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.26.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.26.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.26.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.26.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.26.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.26.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.26.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.26.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.26.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.26.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.26.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.26.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.26.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.26.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.26.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.26.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.26.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.26.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.26.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.27.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.27.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.27.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.27.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.27.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.27.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.27.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.27.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.27.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.27.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.27.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.27.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.27.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.27.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.27.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.27.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.27.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.27.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.27.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.27.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.27.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.27.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.27.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.27.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.27.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.27.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.27.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.27.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.27.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.27.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.27.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.27.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.28.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.28.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.28.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.28.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.28.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.28.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.28.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.28.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.28.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.28.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.28.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.28.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.28.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.28.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.28.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.28.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.28.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.28.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.28.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.28.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.28.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.28.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.28.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.28.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.28.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.28.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.28.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.28.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.28.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.28.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.28.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.28.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.29.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.29.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.29.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.29.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.29.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.29.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.29.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.29.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.29.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.29.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.29.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.29.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.29.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.29.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.29.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.29.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.29.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.29.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.29.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.29.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.29.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.29.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.29.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.29.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.29.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.29.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.29.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.29.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.29.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.29.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.29.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.29.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.3.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.3.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.3.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.3.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.3.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.3.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.3.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.3.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.3.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.3.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.3.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.3.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.3.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.3.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.3.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.3.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.3.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.3.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.3.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.3.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.3.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.3.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.3.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.3.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.3.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.3.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.3.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.3.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.3.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.3.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.3.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.3.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.30.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.30.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.30.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.30.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.30.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.30.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.30.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.30.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.30.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.30.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.30.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.30.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.30.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.30.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.30.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.30.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.30.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.30.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.30.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.30.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.30.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.30.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.30.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.30.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.30.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.30.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.30.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.30.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.30.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.30.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.30.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.30.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.31.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.31.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.31.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.31.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.31.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.31.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.31.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.31.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.31.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.31.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.31.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.31.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.31.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.31.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.31.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.31.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.31.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.31.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.31.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.31.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.31.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.31.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.31.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.31.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.31.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.31.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.31.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.31.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.31.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.31.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.31.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.31.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.32.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.32.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.32.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.32.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.32.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.32.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.32.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.32.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.32.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.32.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.32.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.32.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.32.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.32.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.32.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.32.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.32.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.32.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.32.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.32.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.32.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.32.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.32.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.32.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.32.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.32.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.32.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.32.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.32.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.32.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.32.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.32.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.33.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.33.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.33.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.33.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.33.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.33.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.33.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.33.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.33.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.33.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.33.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.33.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.33.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.33.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.33.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.33.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.33.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.33.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.33.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.33.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.33.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.33.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.33.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.33.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.33.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.33.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.33.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.33.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.33.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.33.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.33.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.33.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.34.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.34.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.34.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.34.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.34.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.34.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.34.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.34.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.34.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.34.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.34.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.34.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.34.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.34.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.34.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.34.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.34.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.34.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.34.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.34.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.34.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.34.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.34.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.34.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.34.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.34.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.34.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.34.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.34.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.34.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.34.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.34.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.35.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.35.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.35.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.35.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.35.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.35.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.35.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.35.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.35.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.35.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.35.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.35.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.35.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.35.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.35.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.35.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.35.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.35.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.35.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.35.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.35.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.35.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.35.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.35.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.35.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.35.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.35.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.35.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.35.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.35.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.35.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.35.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.36.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.36.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.36.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.36.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.36.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.36.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.36.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.36.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.36.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.36.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.36.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.36.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.36.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.36.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.36.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.36.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.36.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.36.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.36.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.36.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.36.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.36.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.36.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.36.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.36.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.36.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.36.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.36.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.36.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.36.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.36.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.36.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.37.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.37.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.37.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.37.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.37.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.37.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.37.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.37.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.37.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.37.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.37.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.37.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.37.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.37.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.37.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.37.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.37.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.37.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.37.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.37.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.37.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.37.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.37.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.37.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.37.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.37.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.37.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.37.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.37.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.37.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.37.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.37.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.38.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.38.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.38.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.38.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.38.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.38.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.38.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.38.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.38.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.38.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.38.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.38.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.38.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.38.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.38.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.38.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.38.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.38.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.38.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.38.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.38.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.38.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.38.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.38.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.38.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.38.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.38.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.38.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.38.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.38.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.38.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.38.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.39.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.39.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.39.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.39.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.39.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.39.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.39.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.39.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.39.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.39.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.39.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.39.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.39.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.39.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.39.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.39.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.39.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.39.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.39.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.39.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.39.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.39.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.39.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.39.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.39.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.39.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.39.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.39.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.39.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.39.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.39.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.39.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.4.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.4.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.4.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.4.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.4.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.4.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.4.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.4.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.4.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.4.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.4.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.4.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.4.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.4.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.4.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.4.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.4.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.4.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.4.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.4.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.4.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.4.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.4.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.4.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.4.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.4.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.4.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.4.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.4.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.4.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.4.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.4.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.40.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.40.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.40.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.40.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.40.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.40.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.40.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.40.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.40.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.40.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.40.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.40.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.40.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.40.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.40.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.40.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.40.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.40.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.40.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.40.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.40.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.40.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.40.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.40.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.40.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.40.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.40.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.40.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.40.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.40.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.40.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.40.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.41.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.41.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.41.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.41.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.41.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.41.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.41.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.41.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.41.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.41.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.41.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.41.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.41.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.41.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.41.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.41.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.41.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.41.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.41.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.41.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.41.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.41.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.41.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.41.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.41.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.41.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.41.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.41.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.41.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.41.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.41.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.41.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.42.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.42.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.42.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.42.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.42.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.42.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.42.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.42.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.42.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.42.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.42.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.42.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.42.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.42.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.42.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.42.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.42.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.42.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.42.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.42.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.42.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.42.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.42.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.42.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.42.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.42.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.42.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.42.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.42.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.42.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.42.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.42.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.43.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.43.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.43.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.43.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.43.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.43.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.43.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.43.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.43.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.43.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.43.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.43.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.43.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.43.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.43.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.43.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.43.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.43.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.43.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.43.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.43.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.43.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.43.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.43.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.43.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.43.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.43.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.43.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.43.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.43.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.43.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.43.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.44.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.44.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.44.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.44.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.44.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.44.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.44.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.44.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.44.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.44.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.44.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.44.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.44.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.44.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.44.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.44.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.44.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.44.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.44.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.44.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.44.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.44.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.44.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.44.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.44.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.44.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.44.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.44.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.44.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.44.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.44.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.44.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.45.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.45.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.45.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.45.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.45.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.45.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.45.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.45.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.45.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.45.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.45.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.45.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.45.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.45.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.45.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.45.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.45.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.45.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.45.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.45.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.45.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.45.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.45.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.45.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.45.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.45.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.45.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.45.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.45.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.45.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.45.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.45.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.46.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.46.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.46.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.46.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.46.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.46.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.46.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.46.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.46.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.46.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.46.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.46.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.46.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.46.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.46.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.46.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.46.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.46.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.46.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.46.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.46.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.46.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.46.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.46.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.46.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.46.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.46.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.46.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.46.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.46.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.46.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.46.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.47.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.47.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.47.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.47.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.47.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.47.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.47.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.47.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.47.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.47.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.47.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.47.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.47.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.47.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.47.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.47.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.47.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.47.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.47.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.47.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.47.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.47.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.47.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.47.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.47.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.47.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.47.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.47.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.47.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.47.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.47.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.47.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.48.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.48.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.48.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.48.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.48.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.48.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.48.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.48.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.48.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.48.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.48.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.48.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.48.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.48.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.48.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.48.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.48.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.48.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.48.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.48.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.48.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.48.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.48.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.48.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.48.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.48.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.48.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.48.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.48.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.48.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.48.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.48.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.49.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.49.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.49.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.49.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.49.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.49.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.49.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.49.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.49.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.49.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.49.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.49.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.49.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.49.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.49.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.49.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.49.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.49.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.49.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.49.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.49.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.49.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.49.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.49.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.49.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.49.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.49.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.49.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.49.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.49.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.49.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.49.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.5.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.5.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.5.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.5.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.5.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.5.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.5.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.5.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.5.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.5.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.5.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.5.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.5.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.5.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.5.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.5.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.5.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.5.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.5.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.5.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.5.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.5.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.5.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.5.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.5.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.5.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.5.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.5.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.5.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.5.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.5.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.5.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.50.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.50.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.50.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.50.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.50.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.50.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.50.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.50.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.50.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.50.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.50.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.50.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.50.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.50.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.50.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.50.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.50.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.50.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.50.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.50.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.50.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.50.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.50.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.50.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.50.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.50.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.50.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.50.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.50.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.50.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.50.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.50.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.51.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.51.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.51.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.51.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.51.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.51.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.51.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.51.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.51.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.51.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.51.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.51.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.51.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.51.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.51.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.51.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.51.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.51.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.51.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.51.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.51.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.51.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.51.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.51.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.51.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.51.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.51.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.51.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.51.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.51.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.51.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.51.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.52.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.52.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.52.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.52.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.52.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.52.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.52.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.52.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.52.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.52.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.52.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.52.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.52.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.52.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.52.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.52.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.52.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.52.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.52.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.52.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.52.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.52.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.52.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.52.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.52.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.52.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.52.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.52.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.52.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.52.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.52.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.52.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.53.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.53.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.53.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.53.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.53.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.53.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.53.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.53.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.53.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.53.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.53.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.53.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.53.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.53.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.53.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.53.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.53.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.53.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.53.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.53.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.53.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.53.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.53.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.53.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.53.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.53.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.53.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.53.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.53.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.53.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.53.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.53.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.54.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.54.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.54.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.54.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.54.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.54.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.54.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.54.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.54.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.54.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.54.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.54.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.54.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.54.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.54.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.54.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.54.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.54.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.54.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.54.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.54.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.54.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.54.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.54.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.54.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.54.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.54.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.54.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.54.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.54.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.54.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.54.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.55.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.55.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.55.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.55.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.55.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.55.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.55.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.55.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.55.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.55.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.55.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.55.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.55.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.55.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.55.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.55.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.55.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.55.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.55.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.55.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.55.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.55.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.55.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.55.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.55.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.55.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.55.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.55.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.55.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.55.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.55.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.55.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.56.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.56.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.56.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.56.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.56.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.56.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.56.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.56.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.56.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.56.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.56.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.56.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.56.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.56.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.56.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.56.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.56.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.56.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.56.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.56.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.56.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.56.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.56.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.56.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.56.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.56.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.56.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.56.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.56.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.56.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.56.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.56.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.57.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.57.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.57.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.57.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.57.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.57.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.57.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.57.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.57.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.57.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.57.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.57.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.57.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.57.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.57.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.57.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.57.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.57.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.57.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.57.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.57.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.57.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.57.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.57.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.57.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.57.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.57.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.57.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.57.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.57.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.57.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.57.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.58.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.58.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.58.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.58.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.58.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.58.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.58.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.58.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.58.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.58.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.58.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.58.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.58.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.58.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.58.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.58.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.58.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.58.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.58.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.58.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.58.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.58.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.58.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.58.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.58.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.58.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.58.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.58.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.58.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.58.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.58.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.58.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.59.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.59.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.59.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.59.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.59.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.59.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.59.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.59.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.59.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.59.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.59.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.59.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.59.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.59.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.59.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.59.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.59.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.59.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.59.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.59.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.59.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.59.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.59.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.59.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.59.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.59.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.59.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.59.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.59.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.59.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.59.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.59.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.6.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.6.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.6.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.6.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.6.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.6.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.6.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.6.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.6.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.6.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.6.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.6.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.6.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.6.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.6.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.6.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.6.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.6.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.6.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.6.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.6.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.6.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.6.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.6.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.6.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.6.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.6.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.6.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.6.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.6.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.6.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.6.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.7.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.7.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.7.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.7.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.7.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.7.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.7.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.7.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.7.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.7.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.7.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.7.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.7.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.7.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.7.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.7.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.7.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.7.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.7.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.7.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.7.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.7.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.7.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.7.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.7.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.7.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.7.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.7.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.7.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.7.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.7.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.7.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.8.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.8.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.8.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.8.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.8.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.8.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.8.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.8.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.8.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.8.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.8.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.8.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.8.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.8.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.8.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.8.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.8.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.8.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.8.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.8.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.8.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.8.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.8.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.8.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.8.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.8.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.8.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.8.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.8.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.8.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.8.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.8.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.9.attn.add_k_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.9.attn.add_k_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.9.attn.add_q_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.9.attn.add_q_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.9.attn.add_v_proj.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.9.attn.add_v_proj.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.9.attn.norm_added_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.9.attn.norm_added_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.9.attn.norm_k.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.9.attn.norm_q.weight,[128],1,torch.bfloat16,BF16,128,16,256
transformer_blocks.9.attn.to_add_out.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.9.attn.to_add_out.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.9.attn.to_k.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.9.attn.to_k.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.9.attn.to_out.0.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.9.attn.to_out.0.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.9.attn.to_q.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.9.attn.to_q.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.9.attn.to_v.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.9.attn.to_v.weight,"[3072, 3072]",2,torch.bfloat16,BF16,9437184,16,18874368
transformer_blocks.9.img_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.9.img_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.9.img_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.9.img_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.9.img_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.9.img_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
transformer_blocks.9.txt_mlp.net.0.proj.bias,[12288],1,torch.bfloat16,BF16,12288,16,24576
transformer_blocks.9.txt_mlp.net.0.proj.weight,"[12288, 3072]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.9.txt_mlp.net.2.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
transformer_blocks.9.txt_mlp.net.2.weight,"[3072, 12288]",2,torch.bfloat16,BF16,37748736,16,75497472
transformer_blocks.9.txt_mod.1.bias,[18432],1,torch.bfloat16,BF16,18432,16,36864
transformer_blocks.9.txt_mod.1.weight,"[18432, 3072]",2,torch.bfloat16,BF16,56623104,16,113246208
txt_in.bias,[3072],1,torch.bfloat16,BF16,3072,16,6144
txt_in.weight,"[3072, 3584]",2,torch.bfloat16,BF16,11010048,16,22020096
txt_norm.weight,[3584],1,torch.bfloat16,BF16,3584,16,7168