{ "denoiser": { "parameter_count": 2061680772, "tensor_count": 652, "tensors": [ { "dtype": "bfloat16", "name": "input_projection.weight", "nbytes": 524288, "numel": 262144, "shape": [ 2048, 128 ] }, { "dtype": "bfloat16", "name": "input_projection.bias", "nbytes": 4096, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "mask_token", "nbytes": 4096, "numel": 2048, "shape": [ 1, 1, 2048 ] }, { "dtype": "bfloat16", "name": "time_embedding.proj_in.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 2048, 256 ] }, { "dtype": "bfloat16", "name": "time_embedding.proj_in.bias", "nbytes": 4096, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "time_embedding.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "time_embedding.proj_out.bias", "nbytes": 4096, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "prefix_layers.0.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "prefix_layers.0.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "prefix_layers.0.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "prefix_layers.0.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.0.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.0.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "prefix_layers.0.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.0.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "prefix_layers.0.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "prefix_layers.0.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "prefix_layers.1.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "prefix_layers.1.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "prefix_layers.1.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "prefix_layers.1.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.1.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.1.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "prefix_layers.1.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.1.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "prefix_layers.1.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "prefix_layers.1.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.0.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.0.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.0.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.0.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.0.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.0.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.0.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.0.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.0.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.0.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.1.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.1.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.1.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.1.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.1.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.1.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.1.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.1.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.1.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.1.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.2.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.2.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.2.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.2.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.2.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.2.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.2.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.2.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.2.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.2.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.3.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.3.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.3.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.3.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.3.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.3.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.3.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.3.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.3.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.3.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.4.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.4.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.4.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.4.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.4.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.4.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.4.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.4.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.4.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.4.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.5.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.5.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.5.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.5.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.5.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.5.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.5.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.5.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.5.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.5.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.6.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.6.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.6.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.6.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.6.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.6.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.6.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.6.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.6.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.6.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.7.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.7.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.7.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.7.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.7.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.7.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.7.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.7.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.7.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.7.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.8.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.8.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.8.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.8.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.8.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.8.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.8.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.8.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.8.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.8.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.9.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.9.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.9.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.9.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.9.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.9.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.9.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.9.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.9.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.9.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.10.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.10.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.10.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.10.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.10.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.10.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.10.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.10.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.10.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.10.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.11.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.11.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.11.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.11.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.11.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.11.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.11.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.11.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.11.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.11.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.12.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.12.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.12.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.12.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.12.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.12.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.12.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.12.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.12.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.12.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.13.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.13.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.13.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.13.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.13.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.13.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.13.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.13.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.13.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.13.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.14.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.14.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.14.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.14.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.14.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.14.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.14.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.14.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.14.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.14.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.15.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.15.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.15.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.15.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.15.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.15.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.15.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.15.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.15.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.15.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.16.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.16.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.16.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.16.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.16.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.16.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.16.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.16.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.16.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.16.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.17.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.17.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.17.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.17.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.17.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.17.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.17.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.17.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.17.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.17.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.18.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.18.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.18.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.18.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.18.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.18.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.18.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.18.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.18.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.18.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.19.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.19.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.19.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.19.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.19.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.19.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.19.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.19.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.19.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.19.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.20.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.20.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.20.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.20.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.20.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.20.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.20.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.20.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.20.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.20.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.21.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.21.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.21.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.21.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.21.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.21.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.21.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.21.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.21.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.21.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.22.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.22.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.22.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.22.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.22.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.22.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.22.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.22.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.22.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.22.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.23.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.23.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.23.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.23.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.23.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.23.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "middle_layers.23.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.23.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.23.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.23.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "suffix_layers.1.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "suffix_layers.1.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "suffix_layers.1.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "suffix_layers.1.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.1.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.1.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "suffix_layers.1.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.1.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "suffix_layers.1.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "suffix_layers.1.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "suffix_layers.2.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "suffix_layers.2.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "suffix_layers.2.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "suffix_layers.2.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.2.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.2.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "suffix_layers.2.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.2.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "suffix_layers.2.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "suffix_layers.2.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "bfloat16", "name": "image_adaln.proj.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "image_adaln.proj.bias", "nbytes": 16384, "numel": 8192, "shape": [ 8192 ] }, { "dtype": "bfloat16", "name": "prefix_layers.0.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.0.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "prefix_layers.1.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.1.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.0.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.0.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.1.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.1.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.2.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.2.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.3.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.3.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.4.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.4.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.5.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.5.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.6.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.6.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.7.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.7.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.8.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.8.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.9.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.9.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.10.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.10.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.11.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.11.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.12.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.12.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.13.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.13.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.14.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.14.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.15.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.15.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.16.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.16.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.17.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.17.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.18.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.18.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.19.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.19.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.20.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.20.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.21.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.21.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.22.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.22.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.23.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.23.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "suffix_layers.1.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.1.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "suffix_layers.2.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.2.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "bfloat16", "name": "fusion.weight", "nbytes": 16777216, "numel": 8388608, "shape": [ 2048, 4096 ] }, { "dtype": "bfloat16", "name": "fusion.bias", "nbytes": 4096, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "tail_scale.weight", "nbytes": 16777216, "numel": 8388608, "shape": [ 4096, 2048 ] }, { "dtype": "bfloat16", "name": "tail_scale.bias", "nbytes": 8192, "numel": 4096, "shape": [ 4096 ] }, { "dtype": "float32", "name": "prefix_layers.2.dit_block.adaln_modulation_scale", "nbytes": 4, "numel": 1, "shape": [] }, { "dtype": "float32", "name": "prefix_layers.2.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "prefix_layers.2.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "prefix_layers.2.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "prefix_layers.2.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.2.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.2.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "prefix_layers.2.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.2.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "prefix_layers.2.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "prefix_layers.2.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "bfloat16", "name": "prefix_layers.2.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.2.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "float32", "name": "prefix_layers.2.cross_modulation_scale", "nbytes": 4, "numel": 1, "shape": [] }, { "dtype": "float32", "name": "prefix_layers.2.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "prefix_layers.2.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "prefix_layers.2.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.2.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "prefix_layers.2.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "prefix_layers.2.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "prefix_layers.2.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "bfloat16", "name": "prefix_layers.2.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "prefix_layers.2.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "float32", "name": "suffix_layers.0.dit_block.adaln_modulation_scale", "nbytes": 4, "numel": 1, "shape": [] }, { "dtype": "float32", "name": "suffix_layers.0.dit_block.attn_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "suffix_layers.0.dit_block.attn_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "suffix_layers.0.dit_block.mlp_norm1.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "suffix_layers.0.dit_block.mlp_norm2.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.0.dit_block.mlp.up.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 8192, 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.0.dit_block.mlp.down.weight", "nbytes": 33554432, "numel": 16777216, "shape": [ 2048, 8192 ] }, { "dtype": "bfloat16", "name": "suffix_layers.0.dit_block.attention.qkv.weight", "nbytes": 25165824, "numel": 12582912, "shape": [ 6144, 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.0.dit_block.attention.proj_out.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "suffix_layers.0.dit_block.attention.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "suffix_layers.0.dit_block.attention.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "bfloat16", "name": "suffix_layers.0.dit_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.0.dit_delta.up.weight", "nbytes": 4194304, "numel": 2097152, "shape": [ 8192, 256 ] }, { "dtype": "float32", "name": "suffix_layers.0.cross_modulation_scale", "nbytes": 4, "numel": 1, "shape": [] }, { "dtype": "float32", "name": "suffix_layers.0.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "suffix_layers.0.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "suffix_layers.0.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.0.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "suffix_layers.0.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "suffix_layers.0.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "suffix_layers.0.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "bfloat16", "name": "suffix_layers.0.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "suffix_layers.0.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "cross_adaln.proj.weight", "nbytes": 16777216, "numel": 8388608, "shape": [ 4096, 2048 ] }, { "dtype": "bfloat16", "name": "cross_adaln.proj.bias", "nbytes": 8192, "numel": 4096, "shape": [ 4096 ] }, { "dtype": "bfloat16", "name": "middle_layers.0.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.0.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.1.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.1.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.2.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.2.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.3.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.3.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.4.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.4.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.5.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.5.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.6.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.6.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.7.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.7.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.8.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.8.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.9.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.9.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.10.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.10.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.11.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.11.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.12.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.12.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.13.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.13.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.14.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.14.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.15.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.15.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.16.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.16.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.17.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.17.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.18.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.18.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.19.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.19.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.20.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.20.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.21.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.21.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.22.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.22.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "bfloat16", "name": "middle_layers.23.cross_delta.down.weight", "nbytes": 1048576, "numel": 524288, "shape": [ 256, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.23.cross_delta.up.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 4096, 256 ] }, { "dtype": "float32", "name": "text_blocks.0.attn_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "float32", "name": "text_blocks.0.mlp_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.0.self_attn.qkv.weight", "nbytes": 6291456, "numel": 3145728, "shape": [ 3072, 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.0.self_attn.proj_out.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 1024, 1024 ] }, { "dtype": "float32", "name": "text_blocks.0.self_attn.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "text_blocks.0.self_attn.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "bfloat16", "name": "text_blocks.0.mlp.up.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.0.mlp.down.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 1024, 4096 ] }, { "dtype": "float32", "name": "text_blocks.1.attn_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "float32", "name": "text_blocks.1.mlp_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.1.self_attn.qkv.weight", "nbytes": 6291456, "numel": 3145728, "shape": [ 3072, 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.1.self_attn.proj_out.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 1024, 1024 ] }, { "dtype": "float32", "name": "text_blocks.1.self_attn.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "text_blocks.1.self_attn.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "bfloat16", "name": "text_blocks.1.mlp.up.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.1.mlp.down.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 1024, 4096 ] }, { "dtype": "float32", "name": "text_blocks.2.attn_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "float32", "name": "text_blocks.2.mlp_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.2.self_attn.qkv.weight", "nbytes": 6291456, "numel": 3145728, "shape": [ 3072, 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.2.self_attn.proj_out.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 1024, 1024 ] }, { "dtype": "float32", "name": "text_blocks.2.self_attn.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "text_blocks.2.self_attn.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "bfloat16", "name": "text_blocks.2.mlp.up.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.2.mlp.down.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 1024, 4096 ] }, { "dtype": "float32", "name": "text_blocks.3.attn_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "float32", "name": "text_blocks.3.mlp_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.3.self_attn.qkv.weight", "nbytes": 6291456, "numel": 3145728, "shape": [ 3072, 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.3.self_attn.proj_out.weight", "nbytes": 2097152, "numel": 1048576, "shape": [ 1024, 1024 ] }, { "dtype": "float32", "name": "text_blocks.3.self_attn.q_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "text_blocks.3.self_attn.k_norm.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "bfloat16", "name": "text_blocks.3.mlp.up.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "text_blocks.3.mlp.down.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 1024, 4096 ] }, { "dtype": "float32", "name": "middle_layers.0.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.0.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.0.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.0.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.0.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.0.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.0.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.1.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.1.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.1.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.1.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.1.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.1.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.1.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.2.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.2.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.2.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.2.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.2.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.2.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.2.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.3.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.3.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.3.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.3.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.3.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.3.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.3.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.4.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.4.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.4.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.4.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.4.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.4.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.4.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.5.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.5.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.5.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.5.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.5.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.5.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.5.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.6.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.6.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.6.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.6.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.6.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.6.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.6.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.7.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.7.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.7.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.7.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.7.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.7.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.7.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.8.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.8.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.8.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.8.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.8.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.8.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.8.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.9.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.9.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.9.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.9.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.9.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.9.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.9.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.10.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.10.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.10.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.10.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.10.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.10.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.10.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.11.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.11.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.11.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.11.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.11.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.11.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.11.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.12.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.12.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.12.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.12.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.12.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.12.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.12.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.13.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.13.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.13.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.13.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.13.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.13.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.13.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.14.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.14.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.14.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.14.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.14.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.14.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.14.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.15.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.15.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.15.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.15.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.15.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.15.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.15.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.16.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.16.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.16.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.16.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.16.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.16.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.16.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.17.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.17.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.17.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.17.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.17.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.17.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.17.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.18.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.18.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.18.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.18.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.18.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.18.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.18.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.19.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.19.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.19.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.19.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.19.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.19.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.19.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.20.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.20.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.20.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.20.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.20.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.20.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.20.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.21.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.21.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.21.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.21.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.21.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.21.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.21.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.22.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.22.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.22.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.22.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.22.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.22.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.22.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.23.cross_attention.image_norm.weight", "nbytes": 8192, "numel": 2048, "shape": [ 2048 ] }, { "dtype": "float32", "name": "middle_layers.23.cross_attention.text_norm.weight", "nbytes": 4096, "numel": 1024, "shape": [ 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.23.cross_attention.q_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "bfloat16", "name": "middle_layers.23.cross_attention.kv_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 4096, 1024 ] }, { "dtype": "bfloat16", "name": "middle_layers.23.cross_attention.out_proj.weight", "nbytes": 8388608, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "middle_layers.23.cross_attention.q_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "middle_layers.23.cross_attention.k_norm_heads.weight", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "output_projection.weight", "nbytes": 1048576, "numel": 262144, "shape": [ 128, 2048 ] }, { "dtype": "float32", "name": "output_projection.bias", "nbytes": 512, "numel": 128, "shape": [ 128 ] }, { "dtype": "float32", "name": "input_position.proj.weight", "nbytes": 16777216, "numel": 4194304, "shape": [ 2048, 2048 ] }, { "dtype": "float32", "name": "unconditional_token", "nbytes": 4096, "numel": 1024, "shape": [ 1, 1024 ] }, { "dtype": "bfloat16", "name": "text_projection_low.weight", "nbytes": 1966080, "numel": 983040, "shape": [ 1024, 960 ] }, { "dtype": "bfloat16", "name": "text_projection_middle.weight", "nbytes": 1966080, "numel": 983040, "shape": [ 1024, 960 ] }, { "dtype": "bfloat16", "name": "text_projection_high.weight", "nbytes": 1966080, "numel": 983040, "shape": [ 1024, 960 ] } ], "total_size": 4132975120 }, "latent_rgb_preview": { "filename": "canter/latent_rgb_preview.safetensors", "parameter_count": 1548, "tensor_count": 2, "tensors": [ { "dtype": "float32", "name": "projection.bias", "nbytes": 48, "numel": 12, "shape": [ 12 ] }, { "dtype": "float32", "name": "projection.weight", "nbytes": 6144, "numel": 1536, "shape": [ 12, 128, 1, 1 ] } ], "total_size": 6192 }, "text_encoder": { "parameter_count": 283161600, "tensor_count": 217, "tensors": [ { "dtype": "float32", "name": "embed_tokens.weight", "nbytes": 188743680, "numel": 47185920, "shape": [ 49152, 960 ] }, { "dtype": "float32", "name": "layers.0.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.0.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.0.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.0.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.0.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.0.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.0.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.0.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.0.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.1.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.1.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.1.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.1.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.1.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.1.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.1.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.1.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.1.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.2.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.2.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.2.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.2.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.2.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.2.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.2.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.2.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.2.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.3.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.3.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.3.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.3.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.3.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.3.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.3.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.3.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.3.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.4.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.4.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.4.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.4.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.4.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.4.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.4.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.4.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.4.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.5.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.5.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.5.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.5.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.5.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.5.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.5.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.5.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.5.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.6.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.6.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.6.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.6.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.6.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.6.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.6.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.6.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.6.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.7.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.7.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.7.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.7.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.7.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.7.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.7.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.7.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.7.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.8.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.8.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.8.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.8.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.8.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.8.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.8.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.8.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.8.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.9.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.9.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.9.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.9.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.9.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.9.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.9.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.9.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.9.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.10.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.10.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.10.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.10.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.10.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.10.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.10.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.10.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.10.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.11.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.11.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.11.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.11.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.11.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.11.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.11.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.11.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.11.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.12.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.12.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.12.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.12.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.12.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.12.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.12.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.12.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.12.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.13.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.13.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.13.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.13.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.13.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.13.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.13.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.13.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.13.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.14.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.14.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.14.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.14.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.14.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.14.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.14.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.14.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.14.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.15.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.15.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.15.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.15.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.15.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.15.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.15.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.15.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.15.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.16.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.16.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.16.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.16.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.16.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.16.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.16.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.16.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.16.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.17.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.17.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.17.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.17.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.17.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.17.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.17.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.17.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.17.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.18.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.18.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.18.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.18.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.18.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.18.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.18.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.18.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.18.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.19.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.19.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.19.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.19.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.19.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.19.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.19.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.19.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.19.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.20.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.20.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.20.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.20.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.20.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.20.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.20.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.20.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.20.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.21.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.21.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.21.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.21.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.21.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.21.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.21.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.21.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.21.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.22.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.22.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.22.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.22.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.22.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.22.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.22.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.22.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.22.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] }, { "dtype": "float32", "name": "layers.23.input_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.23.self_attn.q_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "bfloat16", "name": "layers.23.self_attn.k_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.23.self_attn.v_proj.weight", "nbytes": 614400, "numel": 307200, "shape": [ 320, 960 ] }, { "dtype": "bfloat16", "name": "layers.23.self_attn.o_proj.weight", "nbytes": 1843200, "numel": 921600, "shape": [ 960, 960 ] }, { "dtype": "float32", "name": "layers.23.post_attention_layernorm.weight", "nbytes": 3840, "numel": 960, "shape": [ 960 ] }, { "dtype": "bfloat16", "name": "layers.23.mlp.gate_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.23.mlp.up_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 2560, 960 ] }, { "dtype": "bfloat16", "name": "layers.23.mlp.down_proj.weight", "nbytes": 4915200, "numel": 2457600, "shape": [ 960, 2560 ] } ], "total_size": 660787200 } }