Hugging Face's logo

gatilin
/
unitok_large_visual_codec

Image-to-Image
PyTorch
Safetensors
unitok
image-tokenizer
unified-multimodal
component-extraction
Model card Files
xet
 Community
unitok_large_visual_codec / tensor_mapping.json
gatilin's picture
gatilin
Publish pinned visual components with exact source payloads and native numerical validation
7bfa997 verified
Raw History Blame Contribute Delete
247 kB
{
"logit_scale": {
"source_field": "trainer.unitok.logit_scale",
"shape": [],
"dtype": "torch.float32"
},
"encoder.pos_embed": {
"source_field": "trainer.unitok.encoder.pos_embed",
"shape": [
1,
1,
1024
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stem.conv1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stem.conv1.weight",
"shape": [
160,
3,
3,
3
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stem.conv1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stem.conv1.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stem.norm1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stem.norm1.weight",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stem.norm1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stem.norm1.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stem.conv2.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stem.conv2.weight",
"shape": [
160,
160,
3,
3
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stem.conv2.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stem.conv2.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.0.pre_norm.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.0.pre_norm.weight",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.0.pre_norm.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.0.pre_norm.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.0.conv1_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.0.conv1_1x1.weight",
"shape": [
640,
160,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.0.conv1_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.0.conv1_1x1.bias",
"shape": [
640
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.0.conv2_kxk.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.0.conv2_kxk.weight",
"shape": [
640,
1,
3,
3
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.0.conv2_kxk.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.0.conv2_kxk.bias",
"shape": [
640
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.0.conv3_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.0.conv3_1x1.weight",
"shape": [
160,
640,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.0.conv3_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.0.conv3_1x1.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.1.pre_norm.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.1.pre_norm.weight",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.1.pre_norm.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.1.pre_norm.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.1.conv1_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.1.conv1_1x1.weight",
"shape": [
640,
160,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.1.conv1_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.1.conv1_1x1.bias",
"shape": [
640
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.1.conv2_kxk.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.1.conv2_kxk.weight",
"shape": [
640,
1,
3,
3
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.1.conv2_kxk.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.1.conv2_kxk.bias",
"shape": [
640
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.1.conv3_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.1.conv3_1x1.weight",
"shape": [
160,
640,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.0.1.conv3_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.0.1.conv3_1x1.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.0.shortcut.expand.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.0.shortcut.expand.weight",
"shape": [
320,
160,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.0.shortcut.expand.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.0.shortcut.expand.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.0.pre_norm.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.0.pre_norm.weight",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.0.pre_norm.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.0.pre_norm.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.0.conv1_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.0.conv1_1x1.weight",
"shape": [
1280,
160,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.0.conv1_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.0.conv1_1x1.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.0.conv2_kxk.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.0.conv2_kxk.weight",
"shape": [
1280,
1,
3,
3
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.0.conv2_kxk.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.0.conv2_kxk.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.0.conv3_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.0.conv3_1x1.weight",
"shape": [
320,
1280,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.0.conv3_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.0.conv3_1x1.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.1.pre_norm.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.1.pre_norm.weight",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.1.pre_norm.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.1.pre_norm.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.1.conv1_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.1.conv1_1x1.weight",
"shape": [
1280,
320,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.1.conv1_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.1.conv1_1x1.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.1.conv2_kxk.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.1.conv2_kxk.weight",
"shape": [
1280,
1,
3,
3
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.1.conv2_kxk.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.1.conv2_kxk.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.1.conv3_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.1.conv3_1x1.weight",
"shape": [
320,
1280,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.1.conv3_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.1.conv3_1x1.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.2.pre_norm.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.2.pre_norm.weight",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.2.pre_norm.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.2.pre_norm.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.2.conv1_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.2.conv1_1x1.weight",
"shape": [
1280,
320,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.2.conv1_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.2.conv1_1x1.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.2.conv2_kxk.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.2.conv2_kxk.weight",
"shape": [
1280,
1,
3,
3
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.2.conv2_kxk.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.2.conv2_kxk.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.2.conv3_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.2.conv3_1x1.weight",
"shape": [
320,
1280,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.2.conv3_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.2.conv3_1x1.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.3.pre_norm.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.3.pre_norm.weight",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.3.pre_norm.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.3.pre_norm.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.3.conv1_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.3.conv1_1x1.weight",
"shape": [
1280,
320,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.3.conv1_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.3.conv1_1x1.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.3.conv2_kxk.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.3.conv2_kxk.weight",
"shape": [
1280,
1,
3,
3
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.3.conv2_kxk.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.3.conv2_kxk.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.3.conv3_1x1.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.3.conv3_1x1.weight",
"shape": [
320,
1280,
1,
1
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.stages.1.3.conv3_1x1.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.stages.1.3.conv3_1x1.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.pool.proj.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.pool.proj.weight",
"shape": [
1024,
320,
3,
3
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.pool.proj.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.pool.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.pool.norm.weight": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.pool.norm.weight",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.patch_embed.backbone.pool.norm.bias": {
"source_field": "trainer.unitok.encoder.patch_embed.backbone.pool.norm.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"encoder.blocks.0.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.0.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.0.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.0.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.0.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.0.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.0.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.0.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.0.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.0.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.0.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.0.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.0.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.0.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.0.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.0.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.0.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.0.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.0.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.0.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.0.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.0.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.0.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.0.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.0.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.0.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.0.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.0.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.0.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.0.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.0.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.0.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.1.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.1.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.1.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.1.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.1.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.1.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.1.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.1.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.1.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.1.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.1.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.1.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.1.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.1.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.1.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.1.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.1.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.1.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.1.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.1.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.1.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.2.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.2.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.2.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.2.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.2.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.2.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.2.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.2.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.2.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.2.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.2.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.2.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.2.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.2.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.2.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.2.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.2.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.2.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.2.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.2.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.2.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.3.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.3.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.3.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.3.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.3.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.3.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.3.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.3.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.3.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.3.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.3.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.3.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.3.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.3.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.3.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.3.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.3.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.3.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.3.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.3.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.3.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.4.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.4.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.4.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.4.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.4.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.4.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.4.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.4.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.4.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.4.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.4.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.4.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.4.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.4.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.4.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.4.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.4.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.4.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.4.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.4.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.4.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.5.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.5.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.5.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.5.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.5.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.5.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.5.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.5.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.5.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.5.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.5.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.5.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.5.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.5.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.5.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.5.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.5.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.5.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.5.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.5.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.5.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.6.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.6.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.6.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.6.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.6.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.6.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.6.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.6.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.6.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.6.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.6.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.6.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.6.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.6.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.6.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.6.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.6.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.6.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.6.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.6.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.6.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.7.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.7.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.7.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.7.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.7.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.7.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.7.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.7.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.7.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.7.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.7.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.7.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.7.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.7.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.7.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.7.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.7.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.7.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.7.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.7.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.7.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.8.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.8.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.8.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.8.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.8.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.8.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.8.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.8.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.8.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.8.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.8.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.8.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.8.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.8.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.8.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.8.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.8.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.8.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.8.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.8.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.8.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.9.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.9.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.9.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.9.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.9.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.9.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.9.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.9.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.9.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.9.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.9.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.9.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.9.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.9.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.9.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.9.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.9.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.9.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.9.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.9.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.9.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.10.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.10.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.10.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.10.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.10.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.10.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.10.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.10.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.10.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.10.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.10.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.10.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.10.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.10.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.10.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.10.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.10.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.10.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.10.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.10.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.10.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.11.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.11.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.11.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.11.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.11.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.11.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.11.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.11.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.11.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.11.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.11.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.11.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.11.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.11.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.11.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.11.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.11.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.11.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.11.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.11.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.11.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.12.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.12.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.12.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.12.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.12.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.12.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.12.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.12.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.12.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.12.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.12.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.12.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.12.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.12.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.12.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.12.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.12.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.12.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.12.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.12.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.12.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.13.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.13.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.13.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.13.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.13.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.13.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.13.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.13.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.13.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.13.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.13.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.13.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.13.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.13.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.13.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.13.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.13.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.13.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.13.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.13.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.13.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.14.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.14.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.14.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.14.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.14.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.14.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.14.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.14.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.14.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.14.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.14.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.14.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.14.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.14.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.14.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.14.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.14.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.14.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.14.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.14.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.14.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.15.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.15.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.15.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.15.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.15.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.15.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.15.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.15.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.15.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.15.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.15.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.15.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.15.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.15.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.15.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.15.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.15.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.15.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.15.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.15.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.15.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.16.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.16.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.16.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.16.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.16.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.16.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.16.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.16.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.16.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.16.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.16.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.16.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.16.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.16.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.16.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.16.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.16.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.16.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.16.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.16.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.16.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.17.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.17.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.17.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.17.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.17.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.17.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.17.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.17.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.17.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.17.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.17.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.17.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.17.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.17.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.17.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.17.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.17.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.17.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.17.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.17.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.17.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.18.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.18.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.18.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.18.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.18.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.18.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.18.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.18.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.18.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.18.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.18.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.18.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.18.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.18.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.18.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.18.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.18.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.18.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.18.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.18.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.18.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.19.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.19.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.19.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.19.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.19.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.19.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.19.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.19.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.19.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.19.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.19.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.19.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.19.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.19.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.19.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.19.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.19.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.19.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.19.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.19.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.19.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.20.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.20.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.20.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.20.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.20.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.20.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.20.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.20.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.20.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.20.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.20.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.20.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.20.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.20.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.20.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.20.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.20.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.20.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.20.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.20.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.20.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.21.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.21.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.21.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.21.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.21.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.21.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.21.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.21.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.21.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.21.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.21.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.21.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.21.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.21.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.21.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.21.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.21.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.21.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.21.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.21.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.21.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.22.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.22.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.22.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.22.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.22.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.22.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.22.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.22.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.22.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.22.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.22.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.22.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.22.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.22.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.22.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.22.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.22.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.22.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.22.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.22.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.22.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.23.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.23.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.23.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.23.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.23.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.23.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.23.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.23.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.23.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.23.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.23.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.23.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.23.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.23.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.23.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.23.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.23.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.23.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.23.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.23.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.23.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.24.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.24.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.24.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.24.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.24.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.24.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.24.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.24.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.24.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.24.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.24.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.24.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.24.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.24.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.24.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.24.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.24.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.24.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.24.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.24.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.24.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.25.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.25.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.25.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.25.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.25.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.25.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.25.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.25.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.25.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.25.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.25.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.25.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.25.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.25.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.25.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.25.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.25.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.25.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.25.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.25.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.25.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.26.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.26.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.26.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.26.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.26.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.26.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.26.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.26.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.26.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.26.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.26.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.26.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.26.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.26.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.26.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.26.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.26.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.26.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.26.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.26.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.26.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.27.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.27.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.27.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.27.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.27.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.27.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.27.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.27.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.27.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.27.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.27.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.27.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.27.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.27.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.27.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.27.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.27.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.27.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.27.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.27.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.27.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.28.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.28.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.28.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.28.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.28.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.28.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.28.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.28.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.28.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.28.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.28.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.28.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.28.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.28.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.28.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.28.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.28.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.28.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.28.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.28.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.28.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.29.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.29.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.29.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.29.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.29.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.29.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.29.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.29.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.29.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.29.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.29.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.29.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.29.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.29.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.29.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.29.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.29.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.29.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.29.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.29.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.29.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.norm1.weight": {
"source_field": "trainer.unitok.encoder.blocks.30.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.norm1.bias": {
"source_field": "trainer.unitok.encoder.blocks.30.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.attn.qkv.weight": {
"source_field": "trainer.unitok.encoder.blocks.30.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.attn.qkv.bias": {
"source_field": "trainer.unitok.encoder.blocks.30.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"encoder.blocks.30.attn.proj.weight": {
"source_field": "trainer.unitok.encoder.blocks.30.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.attn.proj.bias": {
"source_field": "trainer.unitok.encoder.blocks.30.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.norm2.weight": {
"source_field": "trainer.unitok.encoder.blocks.30.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.norm2.bias": {
"source_field": "trainer.unitok.encoder.blocks.30.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.mlp.norm.weight": {
"source_field": "trainer.unitok.encoder.blocks.30.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.mlp.norm.bias": {
"source_field": "trainer.unitok.encoder.blocks.30.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.mlp.w0.weight": {
"source_field": "trainer.unitok.encoder.blocks.30.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.mlp.w0.bias": {
"source_field": "trainer.unitok.encoder.blocks.30.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.30.mlp.w1.weight": {
"source_field": "trainer.unitok.encoder.blocks.30.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"encoder.blocks.30.mlp.w1.bias": {
"source_field": "trainer.unitok.encoder.blocks.30.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.30.mlp.w2.weight": {
"source_field": "trainer.unitok.encoder.blocks.30.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"encoder.blocks.30.mlp.w2.bias": {
"source_field": "trainer.unitok.encoder.blocks.30.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.norm.weight": {
"source_field": "trainer.unitok.encoder.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"encoder.norm.bias": {
"source_field": "trainer.unitok.encoder.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"quant_proj.norm1.weight": {
"source_field": "trainer.unitok.quant_proj.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"quant_proj.norm1.bias": {
"source_field": "trainer.unitok.quant_proj.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"quant_proj.attn.q_bias": {
"source_field": "trainer.unitok.quant_proj.attn.q_bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"quant_proj.attn.v_bias": {
"source_field": "trainer.unitok.quant_proj.attn.v_bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"quant_proj.attn.zero_k_bias": {
"source_field": "trainer.unitok.quant_proj.attn.zero_k_bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"quant_proj.attn.qkv.weight": {
"source_field": "trainer.unitok.quant_proj.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"quant_proj.attn.proj.weight": {
"source_field": "trainer.unitok.quant_proj.attn.proj.weight",
"shape": [
64,
64
],
"dtype": "torch.float32"
},
"quant_proj.attn.proj.bias": {
"source_field": "trainer.unitok.quant_proj.attn.proj.bias",
"shape": [
64
],
"dtype": "torch.float32"
},
"quant_proj.proj.weight": {
"source_field": "trainer.unitok.quant_proj.proj.weight",
"shape": [
64,
1024
],
"dtype": "torch.float32"
},
"quant_proj.proj.bias": {
"source_field": "trainer.unitok.quant_proj.proj.bias",
"shape": [
64
],
"dtype": "torch.float32"
},
"quant_proj.norm3.weight": {
"source_field": "trainer.unitok.quant_proj.norm3.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"quant_proj.norm3.bias": {
"source_field": "trainer.unitok.quant_proj.norm3.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"quant_proj.norm2.weight": {
"source_field": "trainer.unitok.quant_proj.norm2.weight",
"shape": [
64
],
"dtype": "torch.float32"
},
"quant_proj.norm2.bias": {
"source_field": "trainer.unitok.quant_proj.norm2.bias",
"shape": [
64
],
"dtype": "torch.float32"
},
"quant_proj.mlp.norm.weight": {
"source_field": "trainer.unitok.quant_proj.mlp.norm.weight",
"shape": [
64
],
"dtype": "torch.float32"
},
"quant_proj.mlp.norm.bias": {
"source_field": "trainer.unitok.quant_proj.mlp.norm.bias",
"shape": [
64
],
"dtype": "torch.float32"
},
"quant_proj.mlp.w0.weight": {
"source_field": "trainer.unitok.quant_proj.mlp.w0.weight",
"shape": [
128,
64
],
"dtype": "torch.float32"
},
"quant_proj.mlp.w0.bias": {
"source_field": "trainer.unitok.quant_proj.mlp.w0.bias",
"shape": [
128
],
"dtype": "torch.float32"
},
"quant_proj.mlp.w1.weight": {
"source_field": "trainer.unitok.quant_proj.mlp.w1.weight",
"shape": [
128,
64
],
"dtype": "torch.float32"
},
"quant_proj.mlp.w1.bias": {
"source_field": "trainer.unitok.quant_proj.mlp.w1.bias",
"shape": [
128
],
"dtype": "torch.float32"
},
"quant_proj.mlp.w2.weight": {
"source_field": "trainer.unitok.quant_proj.mlp.w2.weight",
"shape": [
64,
128
],
"dtype": "torch.float32"
},
"quant_proj.mlp.w2.bias": {
"source_field": "trainer.unitok.quant_proj.mlp.w2.bias",
"shape": [
64
],
"dtype": "torch.float32"
},
"quantizer.codebooks.0.vocab_usage": {
"source_field": "trainer.unitok.quantizer.codebooks.0.vocab_usage",
"shape": [
4096
],
"dtype": "torch.float32"
},
"quantizer.codebooks.0.codebook.weight": {
"source_field": "trainer.unitok.quantizer.codebooks.0.codebook.weight",
"shape": [
4096,
8
],
"dtype": "torch.float32"
},
"quantizer.codebooks.1.vocab_usage": {
"source_field": "trainer.unitok.quantizer.codebooks.1.vocab_usage",
"shape": [
4096
],
"dtype": "torch.float32"
},
"quantizer.codebooks.1.codebook.weight": {
"source_field": "trainer.unitok.quantizer.codebooks.1.codebook.weight",
"shape": [
4096,
8
],
"dtype": "torch.float32"
},
"quantizer.codebooks.2.vocab_usage": {
"source_field": "trainer.unitok.quantizer.codebooks.2.vocab_usage",
"shape": [
4096
],
"dtype": "torch.float32"
},
"quantizer.codebooks.2.codebook.weight": {
"source_field": "trainer.unitok.quantizer.codebooks.2.codebook.weight",
"shape": [
4096,
8
],
"dtype": "torch.float32"
},
"quantizer.codebooks.3.vocab_usage": {
"source_field": "trainer.unitok.quantizer.codebooks.3.vocab_usage",
"shape": [
4096
],
"dtype": "torch.float32"
},
"quantizer.codebooks.3.codebook.weight": {
"source_field": "trainer.unitok.quantizer.codebooks.3.codebook.weight",
"shape": [
4096,
8
],
"dtype": "torch.float32"
},
"quantizer.codebooks.4.vocab_usage": {
"source_field": "trainer.unitok.quantizer.codebooks.4.vocab_usage",
"shape": [
4096
],
"dtype": "torch.float32"
},
"quantizer.codebooks.4.codebook.weight": {
"source_field": "trainer.unitok.quantizer.codebooks.4.codebook.weight",
"shape": [
4096,
8
],
"dtype": "torch.float32"
},
"quantizer.codebooks.5.vocab_usage": {
"source_field": "trainer.unitok.quantizer.codebooks.5.vocab_usage",
"shape": [
4096
],
"dtype": "torch.float32"
},
"quantizer.codebooks.5.codebook.weight": {
"source_field": "trainer.unitok.quantizer.codebooks.5.codebook.weight",
"shape": [
4096,
8
],
"dtype": "torch.float32"
},
"quantizer.codebooks.6.vocab_usage": {
"source_field": "trainer.unitok.quantizer.codebooks.6.vocab_usage",
"shape": [
4096
],
"dtype": "torch.float32"
},
"quantizer.codebooks.6.codebook.weight": {
"source_field": "trainer.unitok.quantizer.codebooks.6.codebook.weight",
"shape": [
4096,
8
],
"dtype": "torch.float32"
},
"quantizer.codebooks.7.vocab_usage": {
"source_field": "trainer.unitok.quantizer.codebooks.7.vocab_usage",
"shape": [
4096
],
"dtype": "torch.float32"
},
"quantizer.codebooks.7.codebook.weight": {
"source_field": "trainer.unitok.quantizer.codebooks.7.codebook.weight",
"shape": [
4096,
8
],
"dtype": "torch.float32"
},
"post_quant_proj.norm1.weight": {
"source_field": "trainer.unitok.post_quant_proj.norm1.weight",
"shape": [
64
],
"dtype": "torch.float32"
},
"post_quant_proj.norm1.bias": {
"source_field": "trainer.unitok.post_quant_proj.norm1.bias",
"shape": [
64
],
"dtype": "torch.float32"
},
"post_quant_proj.attn.q_bias": {
"source_field": "trainer.unitok.post_quant_proj.attn.q_bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.attn.v_bias": {
"source_field": "trainer.unitok.post_quant_proj.attn.v_bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.attn.zero_k_bias": {
"source_field": "trainer.unitok.post_quant_proj.attn.zero_k_bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.attn.qkv.weight": {
"source_field": "trainer.unitok.post_quant_proj.attn.qkv.weight",
"shape": [
3072,
64
],
"dtype": "torch.float32"
},
"post_quant_proj.attn.proj.weight": {
"source_field": "trainer.unitok.post_quant_proj.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.attn.proj.bias": {
"source_field": "trainer.unitok.post_quant_proj.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.proj.weight": {
"source_field": "trainer.unitok.post_quant_proj.proj.weight",
"shape": [
1024,
64
],
"dtype": "torch.float32"
},
"post_quant_proj.proj.bias": {
"source_field": "trainer.unitok.post_quant_proj.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.norm3.weight": {
"source_field": "trainer.unitok.post_quant_proj.norm3.weight",
"shape": [
64
],
"dtype": "torch.float32"
},
"post_quant_proj.norm3.bias": {
"source_field": "trainer.unitok.post_quant_proj.norm3.bias",
"shape": [
64
],
"dtype": "torch.float32"
},
"post_quant_proj.norm2.weight": {
"source_field": "trainer.unitok.post_quant_proj.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.norm2.bias": {
"source_field": "trainer.unitok.post_quant_proj.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.mlp.norm.weight": {
"source_field": "trainer.unitok.post_quant_proj.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.mlp.norm.bias": {
"source_field": "trainer.unitok.post_quant_proj.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.mlp.w0.weight": {
"source_field": "trainer.unitok.post_quant_proj.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.mlp.w0.bias": {
"source_field": "trainer.unitok.post_quant_proj.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"post_quant_proj.mlp.w1.weight": {
"source_field": "trainer.unitok.post_quant_proj.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"post_quant_proj.mlp.w1.bias": {
"source_field": "trainer.unitok.post_quant_proj.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"post_quant_proj.mlp.w2.weight": {
"source_field": "trainer.unitok.post_quant_proj.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"post_quant_proj.mlp.w2.bias": {
"source_field": "trainer.unitok.post_quant_proj.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.0.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.0.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.0.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.0.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.0.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.0.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.0.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.0.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.0.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.0.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.0.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.0.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.0.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.0.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.0.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.0.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.0.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.0.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.0.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.0.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.0.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.1.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.1.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.1.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.1.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.1.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.1.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.1.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.1.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.1.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.1.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.1.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.1.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.1.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.1.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.1.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.1.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.1.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.1.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.1.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.1.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.1.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.2.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.2.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.2.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.2.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.2.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.2.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.2.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.2.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.2.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.2.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.2.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.2.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.2.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.2.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.2.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.2.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.2.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.2.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.2.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.2.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.2.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.3.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.3.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.3.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.3.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.3.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.3.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.3.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.3.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.3.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.3.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.3.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.3.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.3.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.3.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.3.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.3.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.3.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.3.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.3.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.3.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.3.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.4.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.4.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.4.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.4.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.4.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.4.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.4.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.4.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.4.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.4.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.4.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.4.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.4.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.4.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.4.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.4.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.4.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.4.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.4.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.4.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.4.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.5.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.5.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.5.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.5.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.5.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.5.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.5.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.5.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.5.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.5.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.5.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.5.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.5.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.5.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.5.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.5.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.5.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.5.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.5.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.5.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.5.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.6.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.6.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.6.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.6.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.6.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.6.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.6.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.6.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.6.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.6.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.6.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.6.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.6.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.6.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.6.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.6.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.6.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.6.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.6.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.6.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.6.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.7.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.7.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.7.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.7.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.7.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.7.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.7.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.7.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.7.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.7.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.7.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.7.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.7.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.7.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.7.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.7.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.7.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.7.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.7.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.7.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.7.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.8.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.8.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.8.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.8.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.8.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.8.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.8.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.8.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.8.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.8.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.8.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.8.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.8.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.8.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.8.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.8.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.8.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.8.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.8.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.8.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.8.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.9.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.9.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.9.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.9.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.9.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.9.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.9.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.9.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.9.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.9.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.9.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.9.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.9.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.9.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.9.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.9.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.9.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.9.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.9.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.9.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.9.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.10.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.10.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.10.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.10.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.10.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.10.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.10.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.10.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.10.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.10.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.10.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.10.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.10.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.10.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.10.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.10.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.10.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.10.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.10.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.10.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.10.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.11.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.11.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.11.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.11.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.11.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.11.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.11.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.11.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.11.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.11.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.11.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.11.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.11.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.11.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.11.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.11.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.11.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.11.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.11.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.11.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.11.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.12.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.12.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.12.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.12.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.12.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.12.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.12.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.12.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.12.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.12.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.12.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.12.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.12.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.12.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.12.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.12.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.12.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.12.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.12.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.12.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.12.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.13.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.13.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.13.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.13.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.13.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.13.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.13.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.13.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.13.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.13.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.13.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.13.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.13.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.13.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.13.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.13.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.13.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.13.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.13.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.13.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.13.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.14.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.14.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.14.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.14.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.14.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.14.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.14.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.14.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.14.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.14.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.14.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.14.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.14.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.14.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.14.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.14.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.14.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.14.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.14.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.14.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.14.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.15.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.15.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.15.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.15.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.15.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.15.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.15.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.15.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.15.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.15.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.15.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.15.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.15.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.15.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.15.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.15.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.15.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.15.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.15.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.15.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.15.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.16.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.16.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.16.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.16.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.16.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.16.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.16.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.16.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.16.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.16.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.16.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.16.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.16.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.16.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.16.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.16.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.16.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.16.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.16.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.16.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.16.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.17.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.17.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.17.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.17.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.17.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.17.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.17.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.17.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.17.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.17.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.17.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.17.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.17.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.17.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.17.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.17.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.17.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.17.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.17.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.17.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.17.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.18.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.18.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.18.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.18.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.18.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.18.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.18.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.18.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.18.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.18.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.18.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.18.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.18.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.18.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.18.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.18.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.18.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.18.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.18.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.18.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.18.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.19.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.19.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.19.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.19.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.19.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.19.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.19.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.19.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.19.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.19.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.19.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.19.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.19.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.19.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.19.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.19.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.19.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.19.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.19.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.19.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.19.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.20.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.20.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.20.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.20.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.20.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.20.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.20.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.20.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.20.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.20.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.20.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.20.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.20.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.20.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.20.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.20.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.20.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.20.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.20.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.20.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.20.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.21.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.21.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.21.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.21.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.21.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.21.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.21.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.21.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.21.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.21.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.21.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.21.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.21.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.21.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.21.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.21.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.21.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.21.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.21.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.21.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.21.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.22.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.22.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.22.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.22.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.22.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.22.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.22.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.22.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.22.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.22.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.22.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.22.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.22.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.22.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.22.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.22.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.22.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.22.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.22.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.22.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.22.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.23.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.23.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.23.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.23.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.23.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.23.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.23.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.23.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.23.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.23.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.23.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.23.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.23.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.23.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.23.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.23.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.23.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.23.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.23.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.23.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.23.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.24.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.24.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.24.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.24.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.24.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.24.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.24.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.24.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.24.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.24.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.24.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.24.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.24.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.24.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.24.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.24.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.24.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.24.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.24.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.24.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.24.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.25.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.25.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.25.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.25.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.25.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.25.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.25.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.25.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.25.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.25.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.25.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.25.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.25.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.25.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.25.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.25.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.25.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.25.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.25.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.25.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.25.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.26.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.26.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.26.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.26.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.26.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.26.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.26.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.26.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.26.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.26.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.26.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.26.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.26.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.26.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.26.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.26.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.26.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.26.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.26.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.26.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.26.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.27.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.27.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.27.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.27.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.27.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.27.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.27.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.27.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.27.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.27.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.27.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.27.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.27.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.27.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.27.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.27.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.27.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.27.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.27.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.27.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.27.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.28.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.28.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.28.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.28.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.28.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.28.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.28.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.28.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.28.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.28.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.28.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.28.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.28.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.28.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.28.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.28.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.28.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.28.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.28.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.28.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.28.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.29.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.29.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.29.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.29.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.29.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.29.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.29.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.29.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.29.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.29.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.29.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.29.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.29.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.29.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.29.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.29.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.29.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.29.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.29.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.29.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.29.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.norm1.weight": {
"source_field": "trainer.unitok.decoder.blocks.30.norm1.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.norm1.bias": {
"source_field": "trainer.unitok.decoder.blocks.30.norm1.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.attn.qkv.weight": {
"source_field": "trainer.unitok.decoder.blocks.30.attn.qkv.weight",
"shape": [
3072,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.attn.qkv.bias": {
"source_field": "trainer.unitok.decoder.blocks.30.attn.qkv.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"decoder.blocks.30.attn.proj.weight": {
"source_field": "trainer.unitok.decoder.blocks.30.attn.proj.weight",
"shape": [
1024,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.attn.proj.bias": {
"source_field": "trainer.unitok.decoder.blocks.30.attn.proj.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.norm2.weight": {
"source_field": "trainer.unitok.decoder.blocks.30.norm2.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.norm2.bias": {
"source_field": "trainer.unitok.decoder.blocks.30.norm2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.mlp.norm.weight": {
"source_field": "trainer.unitok.decoder.blocks.30.mlp.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.mlp.norm.bias": {
"source_field": "trainer.unitok.decoder.blocks.30.mlp.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.mlp.w0.weight": {
"source_field": "trainer.unitok.decoder.blocks.30.mlp.w0.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.mlp.w0.bias": {
"source_field": "trainer.unitok.decoder.blocks.30.mlp.w0.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.30.mlp.w1.weight": {
"source_field": "trainer.unitok.decoder.blocks.30.mlp.w1.weight",
"shape": [
2048,
1024
],
"dtype": "torch.float32"
},
"decoder.blocks.30.mlp.w1.bias": {
"source_field": "trainer.unitok.decoder.blocks.30.mlp.w1.bias",
"shape": [
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.30.mlp.w2.weight": {
"source_field": "trainer.unitok.decoder.blocks.30.mlp.w2.weight",
"shape": [
1024,
2048
],
"dtype": "torch.float32"
},
"decoder.blocks.30.mlp.w2.bias": {
"source_field": "trainer.unitok.decoder.blocks.30.mlp.w2.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.norm.weight": {
"source_field": "trainer.unitok.decoder.norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.norm.bias": {
"source_field": "trainer.unitok.decoder.norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"decoder.up_conv1.conv.weight": {
"source_field": "trainer.unitok.decoder.up_conv1.conv.weight",
"shape": [
320,
1024,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv1.conv.bias": {
"source_field": "trainer.unitok.decoder.up_conv1.conv.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.shortcut.conv.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.0.shortcut.conv.weight",
"shape": [
320,
320,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.shortcut.conv.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.0.shortcut.conv.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.pre_norm.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.0.pre_norm.weight",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.pre_norm.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.0.pre_norm.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.conv1_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.0.conv1_1x1.weight",
"shape": [
1280,
320,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.conv1_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.0.conv1_1x1.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.up.conv.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.0.up.conv.weight",
"shape": [
1280,
1280,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.up.conv.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.0.up.conv.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.conv2_kxk.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.0.conv2_kxk.weight",
"shape": [
1280,
1,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.conv2_kxk.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.0.conv2_kxk.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.conv3_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.0.conv3_1x1.weight",
"shape": [
320,
1280,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv2.0.conv3_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.0.conv3_1x1.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.1.pre_norm.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.1.pre_norm.weight",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.1.pre_norm.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.1.pre_norm.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.1.conv1_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.1.conv1_1x1.weight",
"shape": [
1280,
320,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv2.1.conv1_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.1.conv1_1x1.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"decoder.up_conv2.1.conv2_kxk.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.1.conv2_kxk.weight",
"shape": [
1280,
1,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv2.1.conv2_kxk.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.1.conv2_kxk.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"decoder.up_conv2.1.conv3_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.1.conv3_1x1.weight",
"shape": [
320,
1280,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv2.1.conv3_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.1.conv3_1x1.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.2.pre_norm.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.2.pre_norm.weight",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.2.pre_norm.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.2.pre_norm.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.2.conv1_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.2.conv1_1x1.weight",
"shape": [
1280,
320,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv2.2.conv1_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.2.conv1_1x1.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"decoder.up_conv2.2.conv2_kxk.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.2.conv2_kxk.weight",
"shape": [
1280,
1,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv2.2.conv2_kxk.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.2.conv2_kxk.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"decoder.up_conv2.2.conv3_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.2.conv3_1x1.weight",
"shape": [
320,
1280,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv2.2.conv3_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.2.conv3_1x1.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.3.pre_norm.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.3.pre_norm.weight",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.3.pre_norm.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.3.pre_norm.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv2.3.conv1_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.3.conv1_1x1.weight",
"shape": [
1280,
320,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv2.3.conv1_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.3.conv1_1x1.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"decoder.up_conv2.3.conv2_kxk.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.3.conv2_kxk.weight",
"shape": [
1280,
1,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv2.3.conv2_kxk.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.3.conv2_kxk.bias",
"shape": [
1280
],
"dtype": "torch.float32"
},
"decoder.up_conv2.3.conv3_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv2.3.conv3_1x1.weight",
"shape": [
320,
1280,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv2.3.conv3_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv2.3.conv3_1x1.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.shortcut.conv.weight": {
"source_field": "trainer.unitok.decoder.up_conv3.0.shortcut.conv.weight",
"shape": [
160,
320,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.shortcut.conv.bias": {
"source_field": "trainer.unitok.decoder.up_conv3.0.shortcut.conv.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.pre_norm.weight": {
"source_field": "trainer.unitok.decoder.up_conv3.0.pre_norm.weight",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.pre_norm.bias": {
"source_field": "trainer.unitok.decoder.up_conv3.0.pre_norm.bias",
"shape": [
320
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.conv1_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv3.0.conv1_1x1.weight",
"shape": [
640,
320,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.conv1_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv3.0.conv1_1x1.bias",
"shape": [
640
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.up.conv.weight": {
"source_field": "trainer.unitok.decoder.up_conv3.0.up.conv.weight",
"shape": [
640,
640,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.up.conv.bias": {
"source_field": "trainer.unitok.decoder.up_conv3.0.up.conv.bias",
"shape": [
640
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.conv2_kxk.weight": {
"source_field": "trainer.unitok.decoder.up_conv3.0.conv2_kxk.weight",
"shape": [
640,
1,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.conv2_kxk.bias": {
"source_field": "trainer.unitok.decoder.up_conv3.0.conv2_kxk.bias",
"shape": [
640
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.conv3_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv3.0.conv3_1x1.weight",
"shape": [
160,
640,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv3.0.conv3_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv3.0.conv3_1x1.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"decoder.up_conv3.1.pre_norm.weight": {
"source_field": "trainer.unitok.decoder.up_conv3.1.pre_norm.weight",
"shape": [
160
],
"dtype": "torch.float32"
},
"decoder.up_conv3.1.pre_norm.bias": {
"source_field": "trainer.unitok.decoder.up_conv3.1.pre_norm.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"decoder.up_conv3.1.conv1_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv3.1.conv1_1x1.weight",
"shape": [
640,
160,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv3.1.conv1_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv3.1.conv1_1x1.bias",
"shape": [
640
],
"dtype": "torch.float32"
},
"decoder.up_conv3.1.conv2_kxk.weight": {
"source_field": "trainer.unitok.decoder.up_conv3.1.conv2_kxk.weight",
"shape": [
640,
1,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv3.1.conv2_kxk.bias": {
"source_field": "trainer.unitok.decoder.up_conv3.1.conv2_kxk.bias",
"shape": [
640
],
"dtype": "torch.float32"
},
"decoder.up_conv3.1.conv3_1x1.weight": {
"source_field": "trainer.unitok.decoder.up_conv3.1.conv3_1x1.weight",
"shape": [
160,
640,
1,
1
],
"dtype": "torch.float32"
},
"decoder.up_conv3.1.conv3_1x1.bias": {
"source_field": "trainer.unitok.decoder.up_conv3.1.conv3_1x1.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"decoder.up_conv4.conv1.conv.weight": {
"source_field": "trainer.unitok.decoder.up_conv4.conv1.conv.weight",
"shape": [
160,
160,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv4.conv1.conv.bias": {
"source_field": "trainer.unitok.decoder.up_conv4.conv1.conv.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"decoder.up_conv4.norm1.weight": {
"source_field": "trainer.unitok.decoder.up_conv4.norm1.weight",
"shape": [
160
],
"dtype": "torch.float32"
},
"decoder.up_conv4.norm1.bias": {
"source_field": "trainer.unitok.decoder.up_conv4.norm1.bias",
"shape": [
160
],
"dtype": "torch.float32"
},
"decoder.up_conv4.conv2.weight": {
"source_field": "trainer.unitok.decoder.up_conv4.conv2.weight",
"shape": [
3,
160,
3,
3
],
"dtype": "torch.float32"
},
"decoder.up_conv4.conv2.bias": {
"source_field": "trainer.unitok.decoder.up_conv4.conv2.bias",
"shape": [
3
],
"dtype": "torch.float32"
},
"text_encoder.text_projection": {
"source_field": "trainer.unitok.text_encoder.text_projection",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.positional_embedding": {
"source_field": "trainer.unitok.text_encoder.positional_embedding",
"shape": [
77,
768
],
"dtype": "torch.float32"
},
"text_encoder.token_embedding.weight": {
"source_field": "trainer.unitok.text_encoder.token_embedding.weight",
"shape": [
49408,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.0.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.0.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.1.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.1.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.2.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.2.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.3.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.3.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.4.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.4.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.5.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.5.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.6.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.6.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.7.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.7.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.8.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.8.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.9.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.9.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.10.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.10.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.ln_1.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.ln_1.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.ln_1.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.ln_1.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.attn.in_proj_weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.attn.in_proj_weight",
"shape": [
2304,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.attn.in_proj_bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.attn.in_proj_bias",
"shape": [
2304
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.attn.out_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.attn.out_proj.weight",
"shape": [
768,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.attn.out_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.attn.out_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.ln_2.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.ln_2.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.ln_2.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.ln_2.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.mlp.c_fc.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.mlp.c_fc.weight",
"shape": [
3072,
768
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.mlp.c_fc.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.mlp.c_fc.bias",
"shape": [
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.mlp.c_proj.weight": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.mlp.c_proj.weight",
"shape": [
768,
3072
],
"dtype": "torch.float32"
},
"text_encoder.transformer.resblocks.11.mlp.c_proj.bias": {
"source_field": "trainer.unitok.text_encoder.transformer.resblocks.11.mlp.c_proj.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.ln_final.weight": {
"source_field": "trainer.unitok.text_encoder.ln_final.weight",
"shape": [
768
],
"dtype": "torch.float32"
},
"text_encoder.ln_final.bias": {
"source_field": "trainer.unitok.text_encoder.ln_final.bias",
"shape": [
768
],
"dtype": "torch.float32"
},
"fc_norm.weight": {
"source_field": "trainer.unitok.fc_norm.weight",
"shape": [
1024
],
"dtype": "torch.float32"
},
"fc_norm.bias": {
"source_field": "trainer.unitok.fc_norm.bias",
"shape": [
1024
],
"dtype": "torch.float32"
},
"projection.weight": {
"source_field": "trainer.unitok.projection.weight",
"shape": [
768,
1024
],
"dtype": "torch.float32"
},
"projection.bias": {
"source_field": "trainer.unitok.projection.bias",
"shape": [
768
],
"dtype": "torch.float32"
}
}