Hugging Face's logo

gatilin
/
showo2_1_5b_visual_flow_components

Feature Extraction
PyTorch
Safetensors
show-o2
unified-multimodal
flow-matching
visual-components
component-extraction
Model card Files
xet
 Community
showo2_1_5b_visual_flow_components / tensor_mapping.json
gatilin's picture
gatilin
Publish pinned visual components with exact source payloads and native numerical validation
b46df1b verified
Raw History Blame Contribute Delete
104 kB
{
"image_embedder_und.proj.weight": {
"source_name": "image_embedder_und.proj.weight",
"shape": [
1152,
16,
2,
2
],
"dtype": "torch.bfloat16"
},
"image_embedder_und.proj.bias": {
"source_name": "image_embedder_und.proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"image_embedder_gen.proj.weight": {
"source_name": "image_embedder_gen.proj.weight",
"shape": [
1536,
16,
2,
2
],
"dtype": "torch.bfloat16"
},
"image_embedder_gen.proj.bias": {
"source_name": "image_embedder_gen.proj.bias",
"shape": [
1536
],
"dtype": "torch.bfloat16"
},
"position_embedding.weight": {
"source_name": "position_embedding.weight",
"shape": [
729,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.0.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.0.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.0.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.0.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.0.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.0.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.0.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.0.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.layer_norm1.weight": {
"source_name": "und_trans.layers.0.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.layer_norm1.bias": {
"source_name": "und_trans.layers.0.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.mlp.fc1.weight": {
"source_name": "und_trans.layers.0.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.mlp.fc1.bias": {
"source_name": "und_trans.layers.0.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.mlp.fc2.weight": {
"source_name": "und_trans.layers.0.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.mlp.fc2.bias": {
"source_name": "und_trans.layers.0.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.layer_norm2.weight": {
"source_name": "und_trans.layers.0.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.0.layer_norm2.bias": {
"source_name": "und_trans.layers.0.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.1.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.1.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.1.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.1.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.1.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.1.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.1.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.1.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.layer_norm1.weight": {
"source_name": "und_trans.layers.1.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.layer_norm1.bias": {
"source_name": "und_trans.layers.1.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.mlp.fc1.weight": {
"source_name": "und_trans.layers.1.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.mlp.fc1.bias": {
"source_name": "und_trans.layers.1.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.mlp.fc2.weight": {
"source_name": "und_trans.layers.1.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.mlp.fc2.bias": {
"source_name": "und_trans.layers.1.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.layer_norm2.weight": {
"source_name": "und_trans.layers.1.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.1.layer_norm2.bias": {
"source_name": "und_trans.layers.1.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.2.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.2.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.2.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.2.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.2.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.2.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.2.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.2.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.layer_norm1.weight": {
"source_name": "und_trans.layers.2.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.layer_norm1.bias": {
"source_name": "und_trans.layers.2.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.mlp.fc1.weight": {
"source_name": "und_trans.layers.2.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.mlp.fc1.bias": {
"source_name": "und_trans.layers.2.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.mlp.fc2.weight": {
"source_name": "und_trans.layers.2.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.mlp.fc2.bias": {
"source_name": "und_trans.layers.2.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.layer_norm2.weight": {
"source_name": "und_trans.layers.2.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.2.layer_norm2.bias": {
"source_name": "und_trans.layers.2.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.3.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.3.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.3.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.3.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.3.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.3.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.3.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.3.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.layer_norm1.weight": {
"source_name": "und_trans.layers.3.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.layer_norm1.bias": {
"source_name": "und_trans.layers.3.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.mlp.fc1.weight": {
"source_name": "und_trans.layers.3.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.mlp.fc1.bias": {
"source_name": "und_trans.layers.3.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.mlp.fc2.weight": {
"source_name": "und_trans.layers.3.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.mlp.fc2.bias": {
"source_name": "und_trans.layers.3.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.layer_norm2.weight": {
"source_name": "und_trans.layers.3.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.3.layer_norm2.bias": {
"source_name": "und_trans.layers.3.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.4.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.4.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.4.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.4.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.4.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.4.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.4.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.4.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.layer_norm1.weight": {
"source_name": "und_trans.layers.4.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.layer_norm1.bias": {
"source_name": "und_trans.layers.4.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.mlp.fc1.weight": {
"source_name": "und_trans.layers.4.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.mlp.fc1.bias": {
"source_name": "und_trans.layers.4.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.mlp.fc2.weight": {
"source_name": "und_trans.layers.4.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.mlp.fc2.bias": {
"source_name": "und_trans.layers.4.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.layer_norm2.weight": {
"source_name": "und_trans.layers.4.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.4.layer_norm2.bias": {
"source_name": "und_trans.layers.4.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.5.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.5.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.5.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.5.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.5.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.5.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.5.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.5.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.layer_norm1.weight": {
"source_name": "und_trans.layers.5.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.layer_norm1.bias": {
"source_name": "und_trans.layers.5.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.mlp.fc1.weight": {
"source_name": "und_trans.layers.5.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.mlp.fc1.bias": {
"source_name": "und_trans.layers.5.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.mlp.fc2.weight": {
"source_name": "und_trans.layers.5.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.mlp.fc2.bias": {
"source_name": "und_trans.layers.5.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.layer_norm2.weight": {
"source_name": "und_trans.layers.5.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.5.layer_norm2.bias": {
"source_name": "und_trans.layers.5.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.6.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.6.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.6.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.6.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.6.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.6.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.6.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.6.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.layer_norm1.weight": {
"source_name": "und_trans.layers.6.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.layer_norm1.bias": {
"source_name": "und_trans.layers.6.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.mlp.fc1.weight": {
"source_name": "und_trans.layers.6.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.mlp.fc1.bias": {
"source_name": "und_trans.layers.6.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.mlp.fc2.weight": {
"source_name": "und_trans.layers.6.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.mlp.fc2.bias": {
"source_name": "und_trans.layers.6.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.layer_norm2.weight": {
"source_name": "und_trans.layers.6.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.6.layer_norm2.bias": {
"source_name": "und_trans.layers.6.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.7.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.7.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.7.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.7.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.7.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.7.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.7.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.7.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.layer_norm1.weight": {
"source_name": "und_trans.layers.7.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.layer_norm1.bias": {
"source_name": "und_trans.layers.7.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.mlp.fc1.weight": {
"source_name": "und_trans.layers.7.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.mlp.fc1.bias": {
"source_name": "und_trans.layers.7.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.mlp.fc2.weight": {
"source_name": "und_trans.layers.7.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.mlp.fc2.bias": {
"source_name": "und_trans.layers.7.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.layer_norm2.weight": {
"source_name": "und_trans.layers.7.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.7.layer_norm2.bias": {
"source_name": "und_trans.layers.7.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.8.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.8.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.8.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.8.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.8.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.8.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.8.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.8.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.layer_norm1.weight": {
"source_name": "und_trans.layers.8.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.layer_norm1.bias": {
"source_name": "und_trans.layers.8.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.mlp.fc1.weight": {
"source_name": "und_trans.layers.8.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.mlp.fc1.bias": {
"source_name": "und_trans.layers.8.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.mlp.fc2.weight": {
"source_name": "und_trans.layers.8.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.mlp.fc2.bias": {
"source_name": "und_trans.layers.8.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.layer_norm2.weight": {
"source_name": "und_trans.layers.8.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.8.layer_norm2.bias": {
"source_name": "und_trans.layers.8.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.9.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.9.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.9.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.9.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.9.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.9.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.9.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.9.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.layer_norm1.weight": {
"source_name": "und_trans.layers.9.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.layer_norm1.bias": {
"source_name": "und_trans.layers.9.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.mlp.fc1.weight": {
"source_name": "und_trans.layers.9.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.mlp.fc1.bias": {
"source_name": "und_trans.layers.9.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.mlp.fc2.weight": {
"source_name": "und_trans.layers.9.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.mlp.fc2.bias": {
"source_name": "und_trans.layers.9.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.layer_norm2.weight": {
"source_name": "und_trans.layers.9.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.9.layer_norm2.bias": {
"source_name": "und_trans.layers.9.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.10.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.10.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.10.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.10.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.10.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.10.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.10.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.10.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.layer_norm1.weight": {
"source_name": "und_trans.layers.10.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.layer_norm1.bias": {
"source_name": "und_trans.layers.10.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.mlp.fc1.weight": {
"source_name": "und_trans.layers.10.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.mlp.fc1.bias": {
"source_name": "und_trans.layers.10.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.mlp.fc2.weight": {
"source_name": "und_trans.layers.10.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.mlp.fc2.bias": {
"source_name": "und_trans.layers.10.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.layer_norm2.weight": {
"source_name": "und_trans.layers.10.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.10.layer_norm2.bias": {
"source_name": "und_trans.layers.10.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.11.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.11.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.11.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.11.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.11.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.11.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.11.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.11.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.layer_norm1.weight": {
"source_name": "und_trans.layers.11.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.layer_norm1.bias": {
"source_name": "und_trans.layers.11.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.mlp.fc1.weight": {
"source_name": "und_trans.layers.11.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.mlp.fc1.bias": {
"source_name": "und_trans.layers.11.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.mlp.fc2.weight": {
"source_name": "und_trans.layers.11.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.mlp.fc2.bias": {
"source_name": "und_trans.layers.11.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.layer_norm2.weight": {
"source_name": "und_trans.layers.11.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.11.layer_norm2.bias": {
"source_name": "und_trans.layers.11.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.12.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.12.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.12.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.12.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.12.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.12.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.12.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.12.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.layer_norm1.weight": {
"source_name": "und_trans.layers.12.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.layer_norm1.bias": {
"source_name": "und_trans.layers.12.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.mlp.fc1.weight": {
"source_name": "und_trans.layers.12.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.mlp.fc1.bias": {
"source_name": "und_trans.layers.12.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.mlp.fc2.weight": {
"source_name": "und_trans.layers.12.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.mlp.fc2.bias": {
"source_name": "und_trans.layers.12.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.layer_norm2.weight": {
"source_name": "und_trans.layers.12.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.12.layer_norm2.bias": {
"source_name": "und_trans.layers.12.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.13.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.13.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.13.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.13.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.13.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.13.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.13.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.13.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.layer_norm1.weight": {
"source_name": "und_trans.layers.13.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.layer_norm1.bias": {
"source_name": "und_trans.layers.13.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.mlp.fc1.weight": {
"source_name": "und_trans.layers.13.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.mlp.fc1.bias": {
"source_name": "und_trans.layers.13.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.mlp.fc2.weight": {
"source_name": "und_trans.layers.13.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.mlp.fc2.bias": {
"source_name": "und_trans.layers.13.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.layer_norm2.weight": {
"source_name": "und_trans.layers.13.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.13.layer_norm2.bias": {
"source_name": "und_trans.layers.13.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.14.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.14.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.14.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.14.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.14.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.14.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.14.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.14.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.layer_norm1.weight": {
"source_name": "und_trans.layers.14.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.layer_norm1.bias": {
"source_name": "und_trans.layers.14.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.mlp.fc1.weight": {
"source_name": "und_trans.layers.14.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.mlp.fc1.bias": {
"source_name": "und_trans.layers.14.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.mlp.fc2.weight": {
"source_name": "und_trans.layers.14.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.mlp.fc2.bias": {
"source_name": "und_trans.layers.14.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.layer_norm2.weight": {
"source_name": "und_trans.layers.14.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.14.layer_norm2.bias": {
"source_name": "und_trans.layers.14.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.15.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.15.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.15.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.15.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.15.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.15.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.15.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.15.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.layer_norm1.weight": {
"source_name": "und_trans.layers.15.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.layer_norm1.bias": {
"source_name": "und_trans.layers.15.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.mlp.fc1.weight": {
"source_name": "und_trans.layers.15.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.mlp.fc1.bias": {
"source_name": "und_trans.layers.15.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.mlp.fc2.weight": {
"source_name": "und_trans.layers.15.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.mlp.fc2.bias": {
"source_name": "und_trans.layers.15.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.layer_norm2.weight": {
"source_name": "und_trans.layers.15.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.15.layer_norm2.bias": {
"source_name": "und_trans.layers.15.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.16.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.16.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.16.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.16.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.16.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.16.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.16.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.16.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.layer_norm1.weight": {
"source_name": "und_trans.layers.16.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.layer_norm1.bias": {
"source_name": "und_trans.layers.16.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.mlp.fc1.weight": {
"source_name": "und_trans.layers.16.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.mlp.fc1.bias": {
"source_name": "und_trans.layers.16.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.mlp.fc2.weight": {
"source_name": "und_trans.layers.16.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.mlp.fc2.bias": {
"source_name": "und_trans.layers.16.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.layer_norm2.weight": {
"source_name": "und_trans.layers.16.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.16.layer_norm2.bias": {
"source_name": "und_trans.layers.16.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.17.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.17.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.17.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.17.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.17.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.17.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.17.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.17.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.layer_norm1.weight": {
"source_name": "und_trans.layers.17.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.layer_norm1.bias": {
"source_name": "und_trans.layers.17.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.mlp.fc1.weight": {
"source_name": "und_trans.layers.17.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.mlp.fc1.bias": {
"source_name": "und_trans.layers.17.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.mlp.fc2.weight": {
"source_name": "und_trans.layers.17.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.mlp.fc2.bias": {
"source_name": "und_trans.layers.17.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.layer_norm2.weight": {
"source_name": "und_trans.layers.17.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.17.layer_norm2.bias": {
"source_name": "und_trans.layers.17.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.18.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.18.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.18.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.18.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.18.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.18.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.18.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.18.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.layer_norm1.weight": {
"source_name": "und_trans.layers.18.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.layer_norm1.bias": {
"source_name": "und_trans.layers.18.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.mlp.fc1.weight": {
"source_name": "und_trans.layers.18.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.mlp.fc1.bias": {
"source_name": "und_trans.layers.18.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.mlp.fc2.weight": {
"source_name": "und_trans.layers.18.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.mlp.fc2.bias": {
"source_name": "und_trans.layers.18.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.layer_norm2.weight": {
"source_name": "und_trans.layers.18.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.18.layer_norm2.bias": {
"source_name": "und_trans.layers.18.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.19.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.19.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.19.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.19.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.19.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.19.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.19.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.19.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.layer_norm1.weight": {
"source_name": "und_trans.layers.19.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.layer_norm1.bias": {
"source_name": "und_trans.layers.19.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.mlp.fc1.weight": {
"source_name": "und_trans.layers.19.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.mlp.fc1.bias": {
"source_name": "und_trans.layers.19.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.mlp.fc2.weight": {
"source_name": "und_trans.layers.19.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.mlp.fc2.bias": {
"source_name": "und_trans.layers.19.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.layer_norm2.weight": {
"source_name": "und_trans.layers.19.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.19.layer_norm2.bias": {
"source_name": "und_trans.layers.19.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.20.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.20.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.20.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.20.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.20.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.20.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.20.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.20.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.layer_norm1.weight": {
"source_name": "und_trans.layers.20.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.layer_norm1.bias": {
"source_name": "und_trans.layers.20.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.mlp.fc1.weight": {
"source_name": "und_trans.layers.20.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.mlp.fc1.bias": {
"source_name": "und_trans.layers.20.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.mlp.fc2.weight": {
"source_name": "und_trans.layers.20.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.mlp.fc2.bias": {
"source_name": "und_trans.layers.20.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.layer_norm2.weight": {
"source_name": "und_trans.layers.20.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.20.layer_norm2.bias": {
"source_name": "und_trans.layers.20.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.21.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.21.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.21.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.21.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.21.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.21.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.21.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.21.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.layer_norm1.weight": {
"source_name": "und_trans.layers.21.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.layer_norm1.bias": {
"source_name": "und_trans.layers.21.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.mlp.fc1.weight": {
"source_name": "und_trans.layers.21.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.mlp.fc1.bias": {
"source_name": "und_trans.layers.21.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.mlp.fc2.weight": {
"source_name": "und_trans.layers.21.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.mlp.fc2.bias": {
"source_name": "und_trans.layers.21.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.layer_norm2.weight": {
"source_name": "und_trans.layers.21.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.21.layer_norm2.bias": {
"source_name": "und_trans.layers.21.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.22.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.22.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.22.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.22.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.22.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.22.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.22.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.22.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.layer_norm1.weight": {
"source_name": "und_trans.layers.22.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.layer_norm1.bias": {
"source_name": "und_trans.layers.22.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.mlp.fc1.weight": {
"source_name": "und_trans.layers.22.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.mlp.fc1.bias": {
"source_name": "und_trans.layers.22.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.mlp.fc2.weight": {
"source_name": "und_trans.layers.22.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.mlp.fc2.bias": {
"source_name": "und_trans.layers.22.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.layer_norm2.weight": {
"source_name": "und_trans.layers.22.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.22.layer_norm2.bias": {
"source_name": "und_trans.layers.22.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.23.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.23.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.23.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.23.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.23.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.23.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.23.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.23.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.layer_norm1.weight": {
"source_name": "und_trans.layers.23.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.layer_norm1.bias": {
"source_name": "und_trans.layers.23.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.mlp.fc1.weight": {
"source_name": "und_trans.layers.23.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.mlp.fc1.bias": {
"source_name": "und_trans.layers.23.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.mlp.fc2.weight": {
"source_name": "und_trans.layers.23.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.mlp.fc2.bias": {
"source_name": "und_trans.layers.23.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.layer_norm2.weight": {
"source_name": "und_trans.layers.23.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.23.layer_norm2.bias": {
"source_name": "und_trans.layers.23.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.24.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.24.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.24.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.24.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.24.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.24.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.24.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.24.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.layer_norm1.weight": {
"source_name": "und_trans.layers.24.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.layer_norm1.bias": {
"source_name": "und_trans.layers.24.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.mlp.fc1.weight": {
"source_name": "und_trans.layers.24.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.mlp.fc1.bias": {
"source_name": "und_trans.layers.24.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.mlp.fc2.weight": {
"source_name": "und_trans.layers.24.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.mlp.fc2.bias": {
"source_name": "und_trans.layers.24.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.layer_norm2.weight": {
"source_name": "und_trans.layers.24.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.24.layer_norm2.bias": {
"source_name": "und_trans.layers.24.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.self_attn.k_proj.weight": {
"source_name": "und_trans.layers.25.self_attn.k_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.self_attn.k_proj.bias": {
"source_name": "und_trans.layers.25.self_attn.k_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.self_attn.v_proj.weight": {
"source_name": "und_trans.layers.25.self_attn.v_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.self_attn.v_proj.bias": {
"source_name": "und_trans.layers.25.self_attn.v_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.self_attn.q_proj.weight": {
"source_name": "und_trans.layers.25.self_attn.q_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.self_attn.q_proj.bias": {
"source_name": "und_trans.layers.25.self_attn.q_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.self_attn.out_proj.weight": {
"source_name": "und_trans.layers.25.self_attn.out_proj.weight",
"shape": [
1152,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.self_attn.out_proj.bias": {
"source_name": "und_trans.layers.25.self_attn.out_proj.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.layer_norm1.weight": {
"source_name": "und_trans.layers.25.layer_norm1.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.layer_norm1.bias": {
"source_name": "und_trans.layers.25.layer_norm1.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.mlp.fc1.weight": {
"source_name": "und_trans.layers.25.mlp.fc1.weight",
"shape": [
4304,
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.mlp.fc1.bias": {
"source_name": "und_trans.layers.25.mlp.fc1.bias",
"shape": [
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.mlp.fc2.weight": {
"source_name": "und_trans.layers.25.mlp.fc2.weight",
"shape": [
1152,
4304
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.mlp.fc2.bias": {
"source_name": "und_trans.layers.25.mlp.fc2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.layer_norm2.weight": {
"source_name": "und_trans.layers.25.layer_norm2.weight",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"und_trans.layers.25.layer_norm2.bias": {
"source_name": "und_trans.layers.25.layer_norm2.bias",
"shape": [
1152
],
"dtype": "torch.bfloat16"
},
"fusion_proj.0.weight": {
"source_name": "fusion_proj.0.weight",
"shape": [
2688
],
"dtype": "torch.bfloat16"
},
"fusion_proj.1.weight": {
"source_name": "fusion_proj.1.weight",
"shape": [
1536,
2688
],
"dtype": "torch.bfloat16"
},
"fusion_proj.1.bias": {
"source_name": "fusion_proj.1.bias",
"shape": [
1536
],
"dtype": "torch.bfloat16"
},
"fusion_proj.3.weight": {
"source_name": "fusion_proj.3.weight",
"shape": [
1536,
1536
],
"dtype": "torch.bfloat16"
},
"fusion_proj.3.bias": {
"source_name": "fusion_proj.3.bias",
"shape": [
1536
],
"dtype": "torch.bfloat16"
},
"time_embed.mlp.0.weight": {
"source_name": "time_embed.mlp.0.weight",
"shape": [
2048,
256
],
"dtype": "torch.bfloat16"
},
"time_embed.mlp.0.bias": {
"source_name": "time_embed.mlp.0.bias",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"time_embed.mlp.2.weight": {
"source_name": "time_embed.mlp.2.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"time_embed.mlp.2.bias": {
"source_name": "time_embed.mlp.2.bias",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diff_proj.0.weight": {
"source_name": "diff_proj.0.weight",
"shape": [
2048,
1536
],
"dtype": "torch.bfloat16"
},
"diff_proj.0.bias": {
"source_name": "diff_proj.0.bias",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diff_proj.2.weight": {
"source_name": "diff_proj.2.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diff_proj.2.bias": {
"source_name": "diff_proj.2.bias",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"time_embed_proj.weight": {
"source_name": "time_embed_proj.weight",
"shape": [
1536,
2048
],
"dtype": "torch.bfloat16"
},
"time_embed_proj.bias": {
"source_name": "time_embed_proj.bias",
"shape": [
1536
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.self_attn.q_proj.weight": {
"source_name": "diffusion_head_a.0.self_attn.q_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.self_attn.k_proj.weight": {
"source_name": "diffusion_head_a.0.self_attn.k_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.self_attn.v_proj.weight": {
"source_name": "diffusion_head_a.0.self_attn.v_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.self_attn.o_proj.weight": {
"source_name": "diffusion_head_a.0.self_attn.o_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.self_attn.q_norm.weight": {
"source_name": "diffusion_head_a.0.self_attn.q_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.self_attn.k_norm.weight": {
"source_name": "diffusion_head_a.0.self_attn.k_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.mlp.gate_proj.weight": {
"source_name": "diffusion_head_a.0.mlp.gate_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.mlp.up_proj.weight": {
"source_name": "diffusion_head_a.0.mlp.up_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.mlp.down_proj.weight": {
"source_name": "diffusion_head_a.0.mlp.down_proj.weight",
"shape": [
2048,
8192
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.input_layernorm.weight": {
"source_name": "diffusion_head_a.0.input_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.post_attention_layernorm.weight": {
"source_name": "diffusion_head_a.0.post_attention_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.adaLN_modulation.1.weight": {
"source_name": "diffusion_head_a.0.adaLN_modulation.1.weight",
"shape": [
12288,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.0.adaLN_modulation.1.bias": {
"source_name": "diffusion_head_a.0.adaLN_modulation.1.bias",
"shape": [
12288
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.self_attn.q_proj.weight": {
"source_name": "diffusion_head_a.1.self_attn.q_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.self_attn.k_proj.weight": {
"source_name": "diffusion_head_a.1.self_attn.k_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.self_attn.v_proj.weight": {
"source_name": "diffusion_head_a.1.self_attn.v_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.self_attn.o_proj.weight": {
"source_name": "diffusion_head_a.1.self_attn.o_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.self_attn.q_norm.weight": {
"source_name": "diffusion_head_a.1.self_attn.q_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.self_attn.k_norm.weight": {
"source_name": "diffusion_head_a.1.self_attn.k_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.mlp.gate_proj.weight": {
"source_name": "diffusion_head_a.1.mlp.gate_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.mlp.up_proj.weight": {
"source_name": "diffusion_head_a.1.mlp.up_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.mlp.down_proj.weight": {
"source_name": "diffusion_head_a.1.mlp.down_proj.weight",
"shape": [
2048,
8192
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.input_layernorm.weight": {
"source_name": "diffusion_head_a.1.input_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.post_attention_layernorm.weight": {
"source_name": "diffusion_head_a.1.post_attention_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.adaLN_modulation.1.weight": {
"source_name": "diffusion_head_a.1.adaLN_modulation.1.weight",
"shape": [
12288,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.1.adaLN_modulation.1.bias": {
"source_name": "diffusion_head_a.1.adaLN_modulation.1.bias",
"shape": [
12288
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.self_attn.q_proj.weight": {
"source_name": "diffusion_head_a.2.self_attn.q_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.self_attn.k_proj.weight": {
"source_name": "diffusion_head_a.2.self_attn.k_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.self_attn.v_proj.weight": {
"source_name": "diffusion_head_a.2.self_attn.v_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.self_attn.o_proj.weight": {
"source_name": "diffusion_head_a.2.self_attn.o_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.self_attn.q_norm.weight": {
"source_name": "diffusion_head_a.2.self_attn.q_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.self_attn.k_norm.weight": {
"source_name": "diffusion_head_a.2.self_attn.k_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.mlp.gate_proj.weight": {
"source_name": "diffusion_head_a.2.mlp.gate_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.mlp.up_proj.weight": {
"source_name": "diffusion_head_a.2.mlp.up_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.mlp.down_proj.weight": {
"source_name": "diffusion_head_a.2.mlp.down_proj.weight",
"shape": [
2048,
8192
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.input_layernorm.weight": {
"source_name": "diffusion_head_a.2.input_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.post_attention_layernorm.weight": {
"source_name": "diffusion_head_a.2.post_attention_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.adaLN_modulation.1.weight": {
"source_name": "diffusion_head_a.2.adaLN_modulation.1.weight",
"shape": [
12288,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.2.adaLN_modulation.1.bias": {
"source_name": "diffusion_head_a.2.adaLN_modulation.1.bias",
"shape": [
12288
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.self_attn.q_proj.weight": {
"source_name": "diffusion_head_a.3.self_attn.q_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.self_attn.k_proj.weight": {
"source_name": "diffusion_head_a.3.self_attn.k_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.self_attn.v_proj.weight": {
"source_name": "diffusion_head_a.3.self_attn.v_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.self_attn.o_proj.weight": {
"source_name": "diffusion_head_a.3.self_attn.o_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.self_attn.q_norm.weight": {
"source_name": "diffusion_head_a.3.self_attn.q_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.self_attn.k_norm.weight": {
"source_name": "diffusion_head_a.3.self_attn.k_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.mlp.gate_proj.weight": {
"source_name": "diffusion_head_a.3.mlp.gate_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.mlp.up_proj.weight": {
"source_name": "diffusion_head_a.3.mlp.up_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.mlp.down_proj.weight": {
"source_name": "diffusion_head_a.3.mlp.down_proj.weight",
"shape": [
2048,
8192
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.input_layernorm.weight": {
"source_name": "diffusion_head_a.3.input_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.post_attention_layernorm.weight": {
"source_name": "diffusion_head_a.3.post_attention_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.adaLN_modulation.1.weight": {
"source_name": "diffusion_head_a.3.adaLN_modulation.1.weight",
"shape": [
12288,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.3.adaLN_modulation.1.bias": {
"source_name": "diffusion_head_a.3.adaLN_modulation.1.bias",
"shape": [
12288
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.self_attn.q_proj.weight": {
"source_name": "diffusion_head_a.4.self_attn.q_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.self_attn.k_proj.weight": {
"source_name": "diffusion_head_a.4.self_attn.k_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.self_attn.v_proj.weight": {
"source_name": "diffusion_head_a.4.self_attn.v_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.self_attn.o_proj.weight": {
"source_name": "diffusion_head_a.4.self_attn.o_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.self_attn.q_norm.weight": {
"source_name": "diffusion_head_a.4.self_attn.q_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.self_attn.k_norm.weight": {
"source_name": "diffusion_head_a.4.self_attn.k_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.mlp.gate_proj.weight": {
"source_name": "diffusion_head_a.4.mlp.gate_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.mlp.up_proj.weight": {
"source_name": "diffusion_head_a.4.mlp.up_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.mlp.down_proj.weight": {
"source_name": "diffusion_head_a.4.mlp.down_proj.weight",
"shape": [
2048,
8192
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.input_layernorm.weight": {
"source_name": "diffusion_head_a.4.input_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.post_attention_layernorm.weight": {
"source_name": "diffusion_head_a.4.post_attention_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.adaLN_modulation.1.weight": {
"source_name": "diffusion_head_a.4.adaLN_modulation.1.weight",
"shape": [
12288,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.4.adaLN_modulation.1.bias": {
"source_name": "diffusion_head_a.4.adaLN_modulation.1.bias",
"shape": [
12288
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.self_attn.q_proj.weight": {
"source_name": "diffusion_head_a.5.self_attn.q_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.self_attn.k_proj.weight": {
"source_name": "diffusion_head_a.5.self_attn.k_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.self_attn.v_proj.weight": {
"source_name": "diffusion_head_a.5.self_attn.v_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.self_attn.o_proj.weight": {
"source_name": "diffusion_head_a.5.self_attn.o_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.self_attn.q_norm.weight": {
"source_name": "diffusion_head_a.5.self_attn.q_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.self_attn.k_norm.weight": {
"source_name": "diffusion_head_a.5.self_attn.k_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.mlp.gate_proj.weight": {
"source_name": "diffusion_head_a.5.mlp.gate_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.mlp.up_proj.weight": {
"source_name": "diffusion_head_a.5.mlp.up_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.mlp.down_proj.weight": {
"source_name": "diffusion_head_a.5.mlp.down_proj.weight",
"shape": [
2048,
8192
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.input_layernorm.weight": {
"source_name": "diffusion_head_a.5.input_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.post_attention_layernorm.weight": {
"source_name": "diffusion_head_a.5.post_attention_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.adaLN_modulation.1.weight": {
"source_name": "diffusion_head_a.5.adaLN_modulation.1.weight",
"shape": [
12288,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.5.adaLN_modulation.1.bias": {
"source_name": "diffusion_head_a.5.adaLN_modulation.1.bias",
"shape": [
12288
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.self_attn.q_proj.weight": {
"source_name": "diffusion_head_a.6.self_attn.q_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.self_attn.k_proj.weight": {
"source_name": "diffusion_head_a.6.self_attn.k_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.self_attn.v_proj.weight": {
"source_name": "diffusion_head_a.6.self_attn.v_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.self_attn.o_proj.weight": {
"source_name": "diffusion_head_a.6.self_attn.o_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.self_attn.q_norm.weight": {
"source_name": "diffusion_head_a.6.self_attn.q_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.self_attn.k_norm.weight": {
"source_name": "diffusion_head_a.6.self_attn.k_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.mlp.gate_proj.weight": {
"source_name": "diffusion_head_a.6.mlp.gate_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.mlp.up_proj.weight": {
"source_name": "diffusion_head_a.6.mlp.up_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.mlp.down_proj.weight": {
"source_name": "diffusion_head_a.6.mlp.down_proj.weight",
"shape": [
2048,
8192
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.input_layernorm.weight": {
"source_name": "diffusion_head_a.6.input_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.post_attention_layernorm.weight": {
"source_name": "diffusion_head_a.6.post_attention_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.adaLN_modulation.1.weight": {
"source_name": "diffusion_head_a.6.adaLN_modulation.1.weight",
"shape": [
12288,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.6.adaLN_modulation.1.bias": {
"source_name": "diffusion_head_a.6.adaLN_modulation.1.bias",
"shape": [
12288
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.self_attn.q_proj.weight": {
"source_name": "diffusion_head_a.7.self_attn.q_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.self_attn.k_proj.weight": {
"source_name": "diffusion_head_a.7.self_attn.k_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.self_attn.v_proj.weight": {
"source_name": "diffusion_head_a.7.self_attn.v_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.self_attn.o_proj.weight": {
"source_name": "diffusion_head_a.7.self_attn.o_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.self_attn.q_norm.weight": {
"source_name": "diffusion_head_a.7.self_attn.q_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.self_attn.k_norm.weight": {
"source_name": "diffusion_head_a.7.self_attn.k_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.mlp.gate_proj.weight": {
"source_name": "diffusion_head_a.7.mlp.gate_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.mlp.up_proj.weight": {
"source_name": "diffusion_head_a.7.mlp.up_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.mlp.down_proj.weight": {
"source_name": "diffusion_head_a.7.mlp.down_proj.weight",
"shape": [
2048,
8192
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.input_layernorm.weight": {
"source_name": "diffusion_head_a.7.input_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.post_attention_layernorm.weight": {
"source_name": "diffusion_head_a.7.post_attention_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.adaLN_modulation.1.weight": {
"source_name": "diffusion_head_a.7.adaLN_modulation.1.weight",
"shape": [
12288,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.7.adaLN_modulation.1.bias": {
"source_name": "diffusion_head_a.7.adaLN_modulation.1.bias",
"shape": [
12288
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.self_attn.q_proj.weight": {
"source_name": "diffusion_head_a.8.self_attn.q_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.self_attn.k_proj.weight": {
"source_name": "diffusion_head_a.8.self_attn.k_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.self_attn.v_proj.weight": {
"source_name": "diffusion_head_a.8.self_attn.v_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.self_attn.o_proj.weight": {
"source_name": "diffusion_head_a.8.self_attn.o_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.self_attn.q_norm.weight": {
"source_name": "diffusion_head_a.8.self_attn.q_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.self_attn.k_norm.weight": {
"source_name": "diffusion_head_a.8.self_attn.k_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.mlp.gate_proj.weight": {
"source_name": "diffusion_head_a.8.mlp.gate_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.mlp.up_proj.weight": {
"source_name": "diffusion_head_a.8.mlp.up_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.mlp.down_proj.weight": {
"source_name": "diffusion_head_a.8.mlp.down_proj.weight",
"shape": [
2048,
8192
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.input_layernorm.weight": {
"source_name": "diffusion_head_a.8.input_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.post_attention_layernorm.weight": {
"source_name": "diffusion_head_a.8.post_attention_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.adaLN_modulation.1.weight": {
"source_name": "diffusion_head_a.8.adaLN_modulation.1.weight",
"shape": [
12288,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.8.adaLN_modulation.1.bias": {
"source_name": "diffusion_head_a.8.adaLN_modulation.1.bias",
"shape": [
12288
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.self_attn.q_proj.weight": {
"source_name": "diffusion_head_a.9.self_attn.q_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.self_attn.k_proj.weight": {
"source_name": "diffusion_head_a.9.self_attn.k_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.self_attn.v_proj.weight": {
"source_name": "diffusion_head_a.9.self_attn.v_proj.weight",
"shape": [
512,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.self_attn.o_proj.weight": {
"source_name": "diffusion_head_a.9.self_attn.o_proj.weight",
"shape": [
2048,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.self_attn.q_norm.weight": {
"source_name": "diffusion_head_a.9.self_attn.q_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.self_attn.k_norm.weight": {
"source_name": "diffusion_head_a.9.self_attn.k_norm.weight",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.mlp.gate_proj.weight": {
"source_name": "diffusion_head_a.9.mlp.gate_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.mlp.up_proj.weight": {
"source_name": "diffusion_head_a.9.mlp.up_proj.weight",
"shape": [
8192,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.mlp.down_proj.weight": {
"source_name": "diffusion_head_a.9.mlp.down_proj.weight",
"shape": [
2048,
8192
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.input_layernorm.weight": {
"source_name": "diffusion_head_a.9.input_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.post_attention_layernorm.weight": {
"source_name": "diffusion_head_a.9.post_attention_layernorm.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.adaLN_modulation.1.weight": {
"source_name": "diffusion_head_a.9.adaLN_modulation.1.weight",
"shape": [
12288,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_a.9.adaLN_modulation.1.bias": {
"source_name": "diffusion_head_a.9.adaLN_modulation.1.bias",
"shape": [
12288
],
"dtype": "torch.bfloat16"
},
"diffusion_head_b.norm_final.weight": {
"source_name": "diffusion_head_b.norm_final.weight",
"shape": [
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_b.linear.weight": {
"source_name": "diffusion_head_b.linear.weight",
"shape": [
64,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_b.linear.bias": {
"source_name": "diffusion_head_b.linear.bias",
"shape": [
64
],
"dtype": "torch.bfloat16"
},
"diffusion_head_b.adaLN_modulation.1.weight": {
"source_name": "diffusion_head_b.adaLN_modulation.1.weight",
"shape": [
4096,
2048
],
"dtype": "torch.bfloat16"
},
"diffusion_head_b.adaLN_modulation.1.bias": {
"source_name": "diffusion_head_b.adaLN_modulation.1.bias",
"shape": [
4096
],
"dtype": "torch.bfloat16"
}
}