Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
wild88
/
granite-vision-4.1
:4b
23
Downloads
Updated
3 days ago
This repository contains models that have been imported from huggingface. The gguf model card here: https://huggingface.co/ibm-granite/granite-vision-4.1-4b-GGUF
This repository contains models that have been imported from huggingface. The gguf model card here: https://huggingface.co/ibm-granite/granite-vision-4.1-4b-GGUF
Cancel
vision
4b
granite-vision-4.1:4b
...
/
projector
573dd2579f60 · 1.2GB
Metadata
general.architecture
clip
clip
general.file_type
F16
F16
clip.has_vision_encoder
true
true
clip.projector_type
granite4_vision
granite4_vision
clip.use_gelu
true
true
clip.vision.attention.head_count
16
16
clip.vision.attention.layer_norm_epsilon
1e-06
1e-06
clip.vision.block_count
27
27
clip.vision.embedding_length
1152
1152
clip.vision.feature_layer
[26, 20, 14, 8, 26, ...]
[26, 20, 14, 8, 26, ...]
clip.vision.feed_forward_length
4304
4304
clip.vision.image_grid_pinpoints
[384, 384, 768, 384, 1152, ...]
[384, 384, 768, 384, 1152, ...]
clip.vision.image_mean
[0.5, 0.5, 0.5]
[0.5, 0.5, 0.5]
clip.vision.image_size
384
384
clip.vision.image_std
[0.5, 0.5, 0.5]
[0.5, 0.5, 0.5]
clip.vision.patch_size
16
16
clip.vision.preproc_image_size
384
384
clip.vision.projection_dim
2560
2560
clip.vision.projector.query_side
4
4
clip.vision.projector.spatial_offsets
[-1, -1, -1, -1, 0, ...]
[-1, -1, -1, -1, 0, ...]
clip.vision.projector.window_side
8
8
Tensor
Name
Type
Shape
v.blk.0
v.blk.0.attn_k.bias
F32
F32
[1152]
v.blk.0.attn_k.weight
F16
F16
[1152, 1152]
v.blk.0.attn_out.bias
F32
F32
[1152]
v.blk.0.attn_out.weight
F16
F16
[1152, 1152]
v.blk.0.attn_q.bias
F32
F32
[1152]
v.blk.0.attn_q.weight
F16
F16
[1152, 1152]
v.blk.0.attn_v.bias
F32
F32
[1152]
v.blk.0.attn_v.weight
F16
F16
[1152, 1152]
v.blk.0.ffn_down.bias
F32
F32
[1152]
v.blk.0.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.0.ffn_up.bias
F32
F32
[4304]
v.blk.0.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.0.ln1.bias
F32
F32
[1152]
v.blk.0.ln1.weight
F32
F32
[1152]
v.blk.0.ln2.bias
F32
F32
[1152]
v.blk.0.ln2.weight
F32
F32
[1152]
v.blk.1
v.blk.1.attn_k.bias
F32
F32
[1152]
v.blk.1.attn_k.weight
F16
F16
[1152, 1152]
v.blk.1.attn_out.bias
F32
F32
[1152]
v.blk.1.attn_out.weight
F16
F16
[1152, 1152]
v.blk.1.attn_q.bias
F32
F32
[1152]
v.blk.1.attn_q.weight
F16
F16
[1152, 1152]
v.blk.1.attn_v.bias
F32
F32
[1152]
v.blk.1.attn_v.weight
F16
F16
[1152, 1152]
v.blk.1.ffn_down.bias
F32
F32
[1152]
v.blk.1.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.1.ffn_up.bias
F32
F32
[4304]
v.blk.1.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.1.ln1.bias
F32
F32
[1152]
v.blk.1.ln1.weight
F32
F32
[1152]
v.blk.1.ln2.bias
F32
F32
[1152]
v.blk.1.ln2.weight
F32
F32
[1152]
v.blk.2
v.blk.2.attn_k.bias
F32
F32
[1152]
v.blk.2.attn_k.weight
F16
F16
[1152, 1152]
v.blk.2.attn_out.bias
F32
F32
[1152]
v.blk.2.attn_out.weight
F16
F16
[1152, 1152]
v.blk.2.attn_q.bias
F32
F32
[1152]
v.blk.2.attn_q.weight
F16
F16
[1152, 1152]
v.blk.2.attn_v.bias
F32
F32
[1152]
v.blk.2.attn_v.weight
F16
F16
[1152, 1152]
v.blk.2.ffn_down.bias
F32
F32
[1152]
v.blk.2.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.2.ffn_up.bias
F32
F32
[4304]
v.blk.2.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.2.ln1.bias
F32
F32
[1152]
v.blk.2.ln1.weight
F32
F32
[1152]
v.blk.2.ln2.bias
F32
F32
[1152]
v.blk.2.ln2.weight
F32
F32
[1152]
v.blk.3
v.blk.3.attn_k.bias
F32
F32
[1152]
v.blk.3.attn_k.weight
F16
F16
[1152, 1152]
v.blk.3.attn_out.bias
F32
F32
[1152]
v.blk.3.attn_out.weight
F16
F16
[1152, 1152]
v.blk.3.attn_q.bias
F32
F32
[1152]
v.blk.3.attn_q.weight
F16
F16
[1152, 1152]
v.blk.3.attn_v.bias
F32
F32
[1152]
v.blk.3.attn_v.weight
F16
F16
[1152, 1152]
v.blk.3.ffn_down.bias
F32
F32
[1152]
v.blk.3.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.3.ffn_up.bias
F32
F32
[4304]
v.blk.3.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.3.ln1.bias
F32
F32
[1152]
v.blk.3.ln1.weight
F32
F32
[1152]
v.blk.3.ln2.bias
F32
F32
[1152]
v.blk.3.ln2.weight
F32
F32
[1152]
v.blk.4
v.blk.4.attn_k.bias
F32
F32
[1152]
v.blk.4.attn_k.weight
F16
F16
[1152, 1152]
v.blk.4.attn_out.bias
F32
F32
[1152]
v.blk.4.attn_out.weight
F16
F16
[1152, 1152]
v.blk.4.attn_q.bias
F32
F32
[1152]
v.blk.4.attn_q.weight
F16
F16
[1152, 1152]
v.blk.4.attn_v.bias
F32
F32
[1152]
v.blk.4.attn_v.weight
F16
F16
[1152, 1152]
v.blk.4.ffn_down.bias
F32
F32
[1152]
v.blk.4.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.4.ffn_up.bias
F32
F32
[4304]
v.blk.4.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.4.ln1.bias
F32
F32
[1152]
v.blk.4.ln1.weight
F32
F32
[1152]
v.blk.4.ln2.bias
F32
F32
[1152]
v.blk.4.ln2.weight
F32
F32
[1152]
v.blk.5
v.blk.5.attn_k.bias
F32
F32
[1152]
v.blk.5.attn_k.weight
F16
F16
[1152, 1152]
v.blk.5.attn_out.bias
F32
F32
[1152]
v.blk.5.attn_out.weight
F16
F16
[1152, 1152]
v.blk.5.attn_q.bias
F32
F32
[1152]
v.blk.5.attn_q.weight
F16
F16
[1152, 1152]
v.blk.5.attn_v.bias
F32
F32
[1152]
v.blk.5.attn_v.weight
F16
F16
[1152, 1152]
v.blk.5.ffn_down.bias
F32
F32
[1152]
v.blk.5.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.5.ffn_up.bias
F32
F32
[4304]
v.blk.5.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.5.ln1.bias
F32
F32
[1152]
v.blk.5.ln1.weight
F32
F32
[1152]
v.blk.5.ln2.bias
F32
F32
[1152]
v.blk.5.ln2.weight
F32
F32
[1152]
v.blk.6
v.blk.6.attn_k.bias
F32
F32
[1152]
v.blk.6.attn_k.weight
F16
F16
[1152, 1152]
v.blk.6.attn_out.bias
F32
F32
[1152]
v.blk.6.attn_out.weight
F16
F16
[1152, 1152]
v.blk.6.attn_q.bias
F32
F32
[1152]
v.blk.6.attn_q.weight
F16
F16
[1152, 1152]
v.blk.6.attn_v.bias
F32
F32
[1152]
v.blk.6.attn_v.weight
F16
F16
[1152, 1152]
v.blk.6.ffn_down.bias
F32
F32
[1152]
v.blk.6.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.6.ffn_up.bias
F32
F32
[4304]
v.blk.6.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.6.ln1.bias
F32
F32
[1152]
v.blk.6.ln1.weight
F32
F32
[1152]
v.blk.6.ln2.bias
F32
F32
[1152]
v.blk.6.ln2.weight
F32
F32
[1152]
v.blk.7
v.blk.7.attn_k.bias
F32
F32
[1152]
v.blk.7.attn_k.weight
F16
F16
[1152, 1152]
v.blk.7.attn_out.bias
F32
F32
[1152]
v.blk.7.attn_out.weight
F16
F16
[1152, 1152]
v.blk.7.attn_q.bias
F32
F32
[1152]
v.blk.7.attn_q.weight
F16
F16
[1152, 1152]
v.blk.7.attn_v.bias
F32
F32
[1152]
v.blk.7.attn_v.weight
F16
F16
[1152, 1152]
v.blk.7.ffn_down.bias
F32
F32
[1152]
v.blk.7.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.7.ffn_up.bias
F32
F32
[4304]
v.blk.7.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.7.ln1.bias
F32
F32
[1152]
v.blk.7.ln1.weight
F32
F32
[1152]
v.blk.7.ln2.bias
F32
F32
[1152]
v.blk.7.ln2.weight
F32
F32
[1152]
v.blk.8
v.blk.8.attn_k.bias
F32
F32
[1152]
v.blk.8.attn_k.weight
F16
F16
[1152, 1152]
v.blk.8.attn_out.bias
F32
F32
[1152]
v.blk.8.attn_out.weight
F16
F16
[1152, 1152]
v.blk.8.attn_q.bias
F32
F32
[1152]
v.blk.8.attn_q.weight
F16
F16
[1152, 1152]
v.blk.8.attn_v.bias
F32
F32
[1152]
v.blk.8.attn_v.weight
F16
F16
[1152, 1152]
v.blk.8.ffn_down.bias
F32
F32
[1152]
v.blk.8.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.8.ffn_up.bias
F32
F32
[4304]
v.blk.8.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.8.ln1.bias
F32
F32
[1152]
v.blk.8.ln1.weight
F32
F32
[1152]
v.blk.8.ln2.bias
F32
F32
[1152]
v.blk.8.ln2.weight
F32
F32
[1152]
v.blk.9
v.blk.9.attn_k.bias
F32
F32
[1152]
v.blk.9.attn_k.weight
F16
F16
[1152, 1152]
v.blk.9.attn_out.bias
F32
F32
[1152]
v.blk.9.attn_out.weight
F16
F16
[1152, 1152]
v.blk.9.attn_q.bias
F32
F32
[1152]
v.blk.9.attn_q.weight
F16
F16
[1152, 1152]
v.blk.9.attn_v.bias
F32
F32
[1152]
v.blk.9.attn_v.weight
F16
F16
[1152, 1152]
v.blk.9.ffn_down.bias
F32
F32
[1152]
v.blk.9.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.9.ffn_up.bias
F32
F32
[4304]
v.blk.9.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.9.ln1.bias
F32
F32
[1152]
v.blk.9.ln1.weight
F32
F32
[1152]
v.blk.9.ln2.bias
F32
F32
[1152]
v.blk.9.ln2.weight
F32
F32
[1152]
v.blk.10
v.blk.10.attn_k.bias
F32
F32
[1152]
v.blk.10.attn_k.weight
F16
F16
[1152, 1152]
v.blk.10.attn_out.bias
F32
F32
[1152]
v.blk.10.attn_out.weight
F16
F16
[1152, 1152]
v.blk.10.attn_q.bias
F32
F32
[1152]
v.blk.10.attn_q.weight
F16
F16
[1152, 1152]
v.blk.10.attn_v.bias
F32
F32
[1152]
v.blk.10.attn_v.weight
F16
F16
[1152, 1152]
v.blk.10.ffn_down.bias
F32
F32
[1152]
v.blk.10.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.10.ffn_up.bias
F32
F32
[4304]
v.blk.10.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.10.ln1.bias
F32
F32
[1152]
v.blk.10.ln1.weight
F32
F32
[1152]
v.blk.10.ln2.bias
F32
F32
[1152]
v.blk.10.ln2.weight
F32
F32
[1152]
v.blk.11
v.blk.11.attn_k.bias
F32
F32
[1152]
v.blk.11.attn_k.weight
F16
F16
[1152, 1152]
v.blk.11.attn_out.bias
F32
F32
[1152]
v.blk.11.attn_out.weight
F16
F16
[1152, 1152]
v.blk.11.attn_q.bias
F32
F32
[1152]
v.blk.11.attn_q.weight
F16
F16
[1152, 1152]
v.blk.11.attn_v.bias
F32
F32
[1152]
v.blk.11.attn_v.weight
F16
F16
[1152, 1152]
v.blk.11.ffn_down.bias
F32
F32
[1152]
v.blk.11.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.11.ffn_up.bias
F32
F32
[4304]
v.blk.11.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.11.ln1.bias
F32
F32
[1152]
v.blk.11.ln1.weight
F32
F32
[1152]
v.blk.11.ln2.bias
F32
F32
[1152]
v.blk.11.ln2.weight
F32
F32
[1152]
v.blk.12
v.blk.12.attn_k.bias
F32
F32
[1152]
v.blk.12.attn_k.weight
F16
F16
[1152, 1152]
v.blk.12.attn_out.bias
F32
F32
[1152]
v.blk.12.attn_out.weight
F16
F16
[1152, 1152]
v.blk.12.attn_q.bias
F32
F32
[1152]
v.blk.12.attn_q.weight
F16
F16
[1152, 1152]
v.blk.12.attn_v.bias
F32
F32
[1152]
v.blk.12.attn_v.weight
F16
F16
[1152, 1152]
v.blk.12.ffn_down.bias
F32
F32
[1152]
v.blk.12.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.12.ffn_up.bias
F32
F32
[4304]
v.blk.12.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.12.ln1.bias
F32
F32
[1152]
v.blk.12.ln1.weight
F32
F32
[1152]
v.blk.12.ln2.bias
F32
F32
[1152]
v.blk.12.ln2.weight
F32
F32
[1152]
v.blk.13
v.blk.13.attn_k.bias
F32
F32
[1152]
v.blk.13.attn_k.weight
F16
F16
[1152, 1152]
v.blk.13.attn_out.bias
F32
F32
[1152]
v.blk.13.attn_out.weight
F16
F16
[1152, 1152]
v.blk.13.attn_q.bias
F32
F32
[1152]
v.blk.13.attn_q.weight
F16
F16
[1152, 1152]
v.blk.13.attn_v.bias
F32
F32
[1152]
v.blk.13.attn_v.weight
F16
F16
[1152, 1152]
v.blk.13.ffn_down.bias
F32
F32
[1152]
v.blk.13.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.13.ffn_up.bias
F32
F32
[4304]
v.blk.13.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.13.ln1.bias
F32
F32
[1152]
v.blk.13.ln1.weight
F32
F32
[1152]
v.blk.13.ln2.bias
F32
F32
[1152]
v.blk.13.ln2.weight
F32
F32
[1152]
v.blk.14
v.blk.14.attn_k.bias
F32
F32
[1152]
v.blk.14.attn_k.weight
F16
F16
[1152, 1152]
v.blk.14.attn_out.bias
F32
F32
[1152]
v.blk.14.attn_out.weight
F16
F16
[1152, 1152]
v.blk.14.attn_q.bias
F32
F32
[1152]
v.blk.14.attn_q.weight
F16
F16
[1152, 1152]
v.blk.14.attn_v.bias
F32
F32
[1152]
v.blk.14.attn_v.weight
F16
F16
[1152, 1152]
v.blk.14.ffn_down.bias
F32
F32
[1152]
v.blk.14.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.14.ffn_up.bias
F32
F32
[4304]
v.blk.14.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.14.ln1.bias
F32
F32
[1152]
v.blk.14.ln1.weight
F32
F32
[1152]
v.blk.14.ln2.bias
F32
F32
[1152]
v.blk.14.ln2.weight
F32
F32
[1152]
v.blk.15
v.blk.15.attn_k.bias
F32
F32
[1152]
v.blk.15.attn_k.weight
F16
F16
[1152, 1152]
v.blk.15.attn_out.bias
F32
F32
[1152]
v.blk.15.attn_out.weight
F16
F16
[1152, 1152]
v.blk.15.attn_q.bias
F32
F32
[1152]
v.blk.15.attn_q.weight
F16
F16
[1152, 1152]
v.blk.15.attn_v.bias
F32
F32
[1152]
v.blk.15.attn_v.weight
F16
F16
[1152, 1152]
v.blk.15.ffn_down.bias
F32
F32
[1152]
v.blk.15.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.15.ffn_up.bias
F32
F32
[4304]
v.blk.15.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.15.ln1.bias
F32
F32
[1152]
v.blk.15.ln1.weight
F32
F32
[1152]
v.blk.15.ln2.bias
F32
F32
[1152]
v.blk.15.ln2.weight
F32
F32
[1152]
v.blk.16
v.blk.16.attn_k.bias
F32
F32
[1152]
v.blk.16.attn_k.weight
F16
F16
[1152, 1152]
v.blk.16.attn_out.bias
F32
F32
[1152]
v.blk.16.attn_out.weight
F16
F16
[1152, 1152]
v.blk.16.attn_q.bias
F32
F32
[1152]
v.blk.16.attn_q.weight
F16
F16
[1152, 1152]
v.blk.16.attn_v.bias
F32
F32
[1152]
v.blk.16.attn_v.weight
F16
F16
[1152, 1152]
v.blk.16.ffn_down.bias
F32
F32
[1152]
v.blk.16.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.16.ffn_up.bias
F32
F32
[4304]
v.blk.16.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.16.ln1.bias
F32
F32
[1152]
v.blk.16.ln1.weight
F32
F32
[1152]
v.blk.16.ln2.bias
F32
F32
[1152]
v.blk.16.ln2.weight
F32
F32
[1152]
v.blk.17
v.blk.17.attn_k.bias
F32
F32
[1152]
v.blk.17.attn_k.weight
F16
F16
[1152, 1152]
v.blk.17.attn_out.bias
F32
F32
[1152]
v.blk.17.attn_out.weight
F16
F16
[1152, 1152]
v.blk.17.attn_q.bias
F32
F32
[1152]
v.blk.17.attn_q.weight
F16
F16
[1152, 1152]
v.blk.17.attn_v.bias
F32
F32
[1152]
v.blk.17.attn_v.weight
F16
F16
[1152, 1152]
v.blk.17.ffn_down.bias
F32
F32
[1152]
v.blk.17.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.17.ffn_up.bias
F32
F32
[4304]
v.blk.17.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.17.ln1.bias
F32
F32
[1152]
v.blk.17.ln1.weight
F32
F32
[1152]
v.blk.17.ln2.bias
F32
F32
[1152]
v.blk.17.ln2.weight
F32
F32
[1152]
v.blk.18
v.blk.18.attn_k.bias
F32
F32
[1152]
v.blk.18.attn_k.weight
F16
F16
[1152, 1152]
v.blk.18.attn_out.bias
F32
F32
[1152]
v.blk.18.attn_out.weight
F16
F16
[1152, 1152]
v.blk.18.attn_q.bias
F32
F32
[1152]
v.blk.18.attn_q.weight
F16
F16
[1152, 1152]
v.blk.18.attn_v.bias
F32
F32
[1152]
v.blk.18.attn_v.weight
F16
F16
[1152, 1152]
v.blk.18.ffn_down.bias
F32
F32
[1152]
v.blk.18.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.18.ffn_up.bias
F32
F32
[4304]
v.blk.18.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.18.ln1.bias
F32
F32
[1152]
v.blk.18.ln1.weight
F32
F32
[1152]
v.blk.18.ln2.bias
F32
F32
[1152]
v.blk.18.ln2.weight
F32
F32
[1152]
v.blk.19
v.blk.19.attn_k.bias
F32
F32
[1152]
v.blk.19.attn_k.weight
F16
F16
[1152, 1152]
v.blk.19.attn_out.bias
F32
F32
[1152]
v.blk.19.attn_out.weight
F16
F16
[1152, 1152]
v.blk.19.attn_q.bias
F32
F32
[1152]
v.blk.19.attn_q.weight
F16
F16
[1152, 1152]
v.blk.19.attn_v.bias
F32
F32
[1152]
v.blk.19.attn_v.weight
F16
F16
[1152, 1152]
v.blk.19.ffn_down.bias
F32
F32
[1152]
v.blk.19.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.19.ffn_up.bias
F32
F32
[4304]
v.blk.19.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.19.ln1.bias
F32
F32
[1152]
v.blk.19.ln1.weight
F32
F32
[1152]
v.blk.19.ln2.bias
F32
F32
[1152]
v.blk.19.ln2.weight
F32
F32
[1152]
v.blk.20
v.blk.20.attn_k.bias
F32
F32
[1152]
v.blk.20.attn_k.weight
F16
F16
[1152, 1152]
v.blk.20.attn_out.bias
F32
F32
[1152]
v.blk.20.attn_out.weight
F16
F16
[1152, 1152]
v.blk.20.attn_q.bias
F32
F32
[1152]
v.blk.20.attn_q.weight
F16
F16
[1152, 1152]
v.blk.20.attn_v.bias
F32
F32
[1152]
v.blk.20.attn_v.weight
F16
F16
[1152, 1152]
v.blk.20.ffn_down.bias
F32
F32
[1152]
v.blk.20.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.20.ffn_up.bias
F32
F32
[4304]
v.blk.20.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.20.ln1.bias
F32
F32
[1152]
v.blk.20.ln1.weight
F32
F32
[1152]
v.blk.20.ln2.bias
F32
F32
[1152]
v.blk.20.ln2.weight
F32
F32
[1152]
v.blk.21
v.blk.21.attn_k.bias
F32
F32
[1152]
v.blk.21.attn_k.weight
F16
F16
[1152, 1152]
v.blk.21.attn_out.bias
F32
F32
[1152]
v.blk.21.attn_out.weight
F16
F16
[1152, 1152]
v.blk.21.attn_q.bias
F32
F32
[1152]
v.blk.21.attn_q.weight
F16
F16
[1152, 1152]
v.blk.21.attn_v.bias
F32
F32
[1152]
v.blk.21.attn_v.weight
F16
F16
[1152, 1152]
v.blk.21.ffn_down.bias
F32
F32
[1152]
v.blk.21.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.21.ffn_up.bias
F32
F32
[4304]
v.blk.21.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.21.ln1.bias
F32
F32
[1152]
v.blk.21.ln1.weight
F32
F32
[1152]
v.blk.21.ln2.bias
F32
F32
[1152]
v.blk.21.ln2.weight
F32
F32
[1152]
v.blk.22
v.blk.22.attn_k.bias
F32
F32
[1152]
v.blk.22.attn_k.weight
F16
F16
[1152, 1152]
v.blk.22.attn_out.bias
F32
F32
[1152]
v.blk.22.attn_out.weight
F16
F16
[1152, 1152]
v.blk.22.attn_q.bias
F32
F32
[1152]
v.blk.22.attn_q.weight
F16
F16
[1152, 1152]
v.blk.22.attn_v.bias
F32
F32
[1152]
v.blk.22.attn_v.weight
F16
F16
[1152, 1152]
v.blk.22.ffn_down.bias
F32
F32
[1152]
v.blk.22.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.22.ffn_up.bias
F32
F32
[4304]
v.blk.22.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.22.ln1.bias
F32
F32
[1152]
v.blk.22.ln1.weight
F32
F32
[1152]
v.blk.22.ln2.bias
F32
F32
[1152]
v.blk.22.ln2.weight
F32
F32
[1152]
v.blk.23
v.blk.23.attn_k.bias
F32
F32
[1152]
v.blk.23.attn_k.weight
F16
F16
[1152, 1152]
v.blk.23.attn_out.bias
F32
F32
[1152]
v.blk.23.attn_out.weight
F16
F16
[1152, 1152]
v.blk.23.attn_q.bias
F32
F32
[1152]
v.blk.23.attn_q.weight
F16
F16
[1152, 1152]
v.blk.23.attn_v.bias
F32
F32
[1152]
v.blk.23.attn_v.weight
F16
F16
[1152, 1152]
v.blk.23.ffn_down.bias
F32
F32
[1152]
v.blk.23.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.23.ffn_up.bias
F32
F32
[4304]
v.blk.23.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.23.ln1.bias
F32
F32
[1152]
v.blk.23.ln1.weight
F32
F32
[1152]
v.blk.23.ln2.bias
F32
F32
[1152]
v.blk.23.ln2.weight
F32
F32
[1152]
v.blk.24
v.blk.24.attn_k.bias
F32
F32
[1152]
v.blk.24.attn_k.weight
F16
F16
[1152, 1152]
v.blk.24.attn_out.bias
F32
F32
[1152]
v.blk.24.attn_out.weight
F16
F16
[1152, 1152]
v.blk.24.attn_q.bias
F32
F32
[1152]
v.blk.24.attn_q.weight
F16
F16
[1152, 1152]
v.blk.24.attn_v.bias
F32
F32
[1152]
v.blk.24.attn_v.weight
F16
F16
[1152, 1152]
v.blk.24.ffn_down.bias
F32
F32
[1152]
v.blk.24.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.24.ffn_up.bias
F32
F32
[4304]
v.blk.24.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.24.ln1.bias
F32
F32
[1152]
v.blk.24.ln1.weight
F32
F32
[1152]
v.blk.24.ln2.bias
F32
F32
[1152]
v.blk.24.ln2.weight
F32
F32
[1152]
v.blk.25
v.blk.25.attn_k.bias
F32
F32
[1152]
v.blk.25.attn_k.weight
F16
F16
[1152, 1152]
v.blk.25.attn_out.bias
F32
F32
[1152]
v.blk.25.attn_out.weight
F16
F16
[1152, 1152]
v.blk.25.attn_q.bias
F32
F32
[1152]
v.blk.25.attn_q.weight
F16
F16
[1152, 1152]
v.blk.25.attn_v.bias
F32
F32
[1152]
v.blk.25.attn_v.weight
F16
F16
[1152, 1152]
v.blk.25.ffn_down.bias
F32
F32
[1152]
v.blk.25.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.25.ffn_up.bias
F32
F32
[4304]
v.blk.25.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.25.ln1.bias
F32
F32
[1152]
v.blk.25.ln1.weight
F32
F32
[1152]
v.blk.25.ln2.bias
F32
F32
[1152]
v.blk.25.ln2.weight
F32
F32
[1152]
v.blk.26
v.blk.26.attn_k.bias
F32
F32
[1152]
v.blk.26.attn_k.weight
F16
F16
[1152, 1152]
v.blk.26.attn_out.bias
F32
F32
[1152]
v.blk.26.attn_out.weight
F16
F16
[1152, 1152]
v.blk.26.attn_q.bias
F32
F32
[1152]
v.blk.26.attn_q.weight
F16
F16
[1152, 1152]
v.blk.26.attn_v.bias
F32
F32
[1152]
v.blk.26.attn_v.weight
F16
F16
[1152, 1152]
v.blk.26.ffn_down.bias
F32
F32
[1152]
v.blk.26.ffn_down.weight
F16
F16
[4304, 1152]
v.blk.26.ffn_up.bias
F32
F32
[4304]
v.blk.26.ffn_up.weight
F16
F16
[1152, 4304]
v.blk.26.ln1.bias
F32
F32
[1152]
v.blk.26.ln1.weight
F32
F32
[1152]
v.blk.26.ln2.bias
F32
F32
[1152]
v.blk.26.ln2.weight
F32
F32
[1152]
v.image_newline
F32
F32
[2560]
v.patch_embd.bias
F32
F32
[1152]
v.patch_embd.weight
F16
F16
[16, 16, 3, 1152]
v.position_embd.weight
F32
F32
[1152, 576]
v.post_ln.bias
F32
F32
[1152]
v.post_ln.weight
F32
F32
[1152]
v.proj_blk.0.cross_attn_k.bias
F32
F32
[1152]
v.proj_blk.0.cross_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.0.cross_attn_norm.bias
F32
F32
[1152]
v.proj_blk.0.cross_attn_norm.weight
F32
F32
[1152]
v.proj_blk.0.cross_attn_out.bias
F32
F32
[1152]
v.proj_blk.0.cross_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.0.cross_attn_q.bias
F32
F32
[1152]
v.proj_blk.0.cross_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.0.cross_attn_v.bias
F32
F32
[1152]
v.proj_blk.0.cross_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.0.ffn_down.bias
F32
F32
[1152]
v.proj_blk.0.ffn_down.weight
F16
F16
[3072, 1152]
v.proj_blk.0.ffn_norm.bias
F32
F32
[1152]
v.proj_blk.0.ffn_norm.weight
F32
F32
[1152]
v.proj_blk.0.ffn_up.bias
F32
F32
[3072]
v.proj_blk.0.ffn_up.weight
F16
F16
[1152, 3072]
v.proj_blk.0.img_pos
F32
F32
[1152, 64, 1]
v.proj_blk.0.linear.bias
F32
F32
[2560]
v.proj_blk.0.linear.weight
F16
F16
[1152, 2560]
v.proj_blk.0.norm.bias
F32
F32
[1152]
v.proj_blk.0.norm.weight
F32
F32
[1152]
v.proj_blk.0.post_norm.bias
F32
F32
[1152]
v.proj_blk.0.post_norm.weight
F32
F32
[1152]
v.proj_blk.0.query
F32
F32
[1152, 16, 1]
v.proj_blk.0.self_attn_k.bias
F32
F32
[1152]
v.proj_blk.0.self_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.0.self_attn_norm.bias
F32
F32
[1152]
v.proj_blk.0.self_attn_norm.weight
F32
F32
[1152]
v.proj_blk.0.self_attn_out.bias
F32
F32
[1152]
v.proj_blk.0.self_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.0.self_attn_q.bias
F32
F32
[1152]
v.proj_blk.0.self_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.0.self_attn_v.bias
F32
F32
[1152]
v.proj_blk.0.self_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.1.cross_attn_k.bias
F32
F32
[1152]
v.proj_blk.1.cross_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.1.cross_attn_norm.bias
F32
F32
[1152]
v.proj_blk.1.cross_attn_norm.weight
F32
F32
[1152]
v.proj_blk.1.cross_attn_out.bias
F32
F32
[1152]
v.proj_blk.1.cross_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.1.cross_attn_q.bias
F32
F32
[1152]
v.proj_blk.1.cross_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.1.cross_attn_v.bias
F32
F32
[1152]
v.proj_blk.1.cross_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.1.ffn_down.bias
F32
F32
[1152]
v.proj_blk.1.ffn_down.weight
F16
F16
[3072, 1152]
v.proj_blk.1.ffn_norm.bias
F32
F32
[1152]
v.proj_blk.1.ffn_norm.weight
F32
F32
[1152]
v.proj_blk.1.ffn_up.bias
F32
F32
[3072]
v.proj_blk.1.ffn_up.weight
F16
F16
[1152, 3072]
v.proj_blk.1.img_pos
F32
F32
[1152, 64, 1]
v.proj_blk.1.linear.bias
F32
F32
[2560]
v.proj_blk.1.linear.weight
F16
F16
[1152, 2560]
v.proj_blk.1.norm.bias
F32
F32
[1152]
v.proj_blk.1.norm.weight
F32
F32
[1152]
v.proj_blk.1.post_norm.bias
F32
F32
[1152]
v.proj_blk.1.post_norm.weight
F32
F32
[1152]
v.proj_blk.1.query
F32
F32
[1152, 16, 1]
v.proj_blk.1.self_attn_k.bias
F32
F32
[1152]
v.proj_blk.1.self_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.1.self_attn_norm.bias
F32
F32
[1152]
v.proj_blk.1.self_attn_norm.weight
F32
F32
[1152]
v.proj_blk.1.self_attn_out.bias
F32
F32
[1152]
v.proj_blk.1.self_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.1.self_attn_q.bias
F32
F32
[1152]
v.proj_blk.1.self_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.1.self_attn_v.bias
F32
F32
[1152]
v.proj_blk.1.self_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.2.cross_attn_k.bias
F32
F32
[1152]
v.proj_blk.2.cross_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.2.cross_attn_norm.bias
F32
F32
[1152]
v.proj_blk.2.cross_attn_norm.weight
F32
F32
[1152]
v.proj_blk.2.cross_attn_out.bias
F32
F32
[1152]
v.proj_blk.2.cross_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.2.cross_attn_q.bias
F32
F32
[1152]
v.proj_blk.2.cross_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.2.cross_attn_v.bias
F32
F32
[1152]
v.proj_blk.2.cross_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.2.ffn_down.bias
F32
F32
[1152]
v.proj_blk.2.ffn_down.weight
F16
F16
[3072, 1152]
v.proj_blk.2.ffn_norm.bias
F32
F32
[1152]
v.proj_blk.2.ffn_norm.weight
F32
F32
[1152]
v.proj_blk.2.ffn_up.bias
F32
F32
[3072]
v.proj_blk.2.ffn_up.weight
F16
F16
[1152, 3072]
v.proj_blk.2.img_pos
F32
F32
[1152, 64, 1]
v.proj_blk.2.linear.bias
F32
F32
[2560]
v.proj_blk.2.linear.weight
F16
F16
[1152, 2560]
v.proj_blk.2.norm.bias
F32
F32
[1152]
v.proj_blk.2.norm.weight
F32
F32
[1152]
v.proj_blk.2.post_norm.bias
F32
F32
[1152]
v.proj_blk.2.post_norm.weight
F32
F32
[1152]
v.proj_blk.2.query
F32
F32
[1152, 16, 1]
v.proj_blk.2.self_attn_k.bias
F32
F32
[1152]
v.proj_blk.2.self_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.2.self_attn_norm.bias
F32
F32
[1152]
v.proj_blk.2.self_attn_norm.weight
F32
F32
[1152]
v.proj_blk.2.self_attn_out.bias
F32
F32
[1152]
v.proj_blk.2.self_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.2.self_attn_q.bias
F32
F32
[1152]
v.proj_blk.2.self_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.2.self_attn_v.bias
F32
F32
[1152]
v.proj_blk.2.self_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.3.cross_attn_k.bias
F32
F32
[1152]
v.proj_blk.3.cross_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.3.cross_attn_norm.bias
F32
F32
[1152]
v.proj_blk.3.cross_attn_norm.weight
F32
F32
[1152]
v.proj_blk.3.cross_attn_out.bias
F32
F32
[1152]
v.proj_blk.3.cross_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.3.cross_attn_q.bias
F32
F32
[1152]
v.proj_blk.3.cross_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.3.cross_attn_v.bias
F32
F32
[1152]
v.proj_blk.3.cross_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.3.ffn_down.bias
F32
F32
[1152]
v.proj_blk.3.ffn_down.weight
F16
F16
[3072, 1152]
v.proj_blk.3.ffn_norm.bias
F32
F32
[1152]
v.proj_blk.3.ffn_norm.weight
F32
F32
[1152]
v.proj_blk.3.ffn_up.bias
F32
F32
[3072]
v.proj_blk.3.ffn_up.weight
F16
F16
[1152, 3072]
v.proj_blk.3.img_pos
F32
F32
[1152, 64, 1]
v.proj_blk.3.linear.bias
F32
F32
[2560]
v.proj_blk.3.linear.weight
F16
F16
[1152, 2560]
v.proj_blk.3.norm.bias
F32
F32
[1152]
v.proj_blk.3.norm.weight
F32
F32
[1152]
v.proj_blk.3.post_norm.bias
F32
F32
[1152]
v.proj_blk.3.post_norm.weight
F32
F32
[1152]
v.proj_blk.3.query
F32
F32
[1152, 16, 1]
v.proj_blk.3.self_attn_k.bias
F32
F32
[1152]
v.proj_blk.3.self_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.3.self_attn_norm.bias
F32
F32
[1152]
v.proj_blk.3.self_attn_norm.weight
F32
F32
[1152]
v.proj_blk.3.self_attn_out.bias
F32
F32
[1152]
v.proj_blk.3.self_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.3.self_attn_q.bias
F32
F32
[1152]
v.proj_blk.3.self_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.3.self_attn_v.bias
F32
F32
[1152]
v.proj_blk.3.self_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.4.cross_attn_k.bias
F32
F32
[1152]
v.proj_blk.4.cross_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.4.cross_attn_norm.bias
F32
F32
[1152]
v.proj_blk.4.cross_attn_norm.weight
F32
F32
[1152]
v.proj_blk.4.cross_attn_out.bias
F32
F32
[1152]
v.proj_blk.4.cross_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.4.cross_attn_q.bias
F32
F32
[1152]
v.proj_blk.4.cross_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.4.cross_attn_v.bias
F32
F32
[1152]
v.proj_blk.4.cross_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.4.ffn_down.bias
F32
F32
[1152]
v.proj_blk.4.ffn_down.weight
F16
F16
[3072, 1152]
v.proj_blk.4.ffn_norm.bias
F32
F32
[1152]
v.proj_blk.4.ffn_norm.weight
F32
F32
[1152]
v.proj_blk.4.ffn_up.bias
F32
F32
[3072]
v.proj_blk.4.ffn_up.weight
F16
F16
[1152, 3072]
v.proj_blk.4.img_pos
F32
F32
[1152, 64, 1]
v.proj_blk.4.linear.bias
F32
F32
[2560]
v.proj_blk.4.linear.weight
F16
F16
[1152, 2560]
v.proj_blk.4.norm.bias
F32
F32
[1152]
v.proj_blk.4.norm.weight
F32
F32
[1152]
v.proj_blk.4.post_norm.bias
F32
F32
[1152]
v.proj_blk.4.post_norm.weight
F32
F32
[1152]
v.proj_blk.4.query
F32
F32
[1152, 16, 1]
v.proj_blk.4.self_attn_k.bias
F32
F32
[1152]
v.proj_blk.4.self_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.4.self_attn_norm.bias
F32
F32
[1152]
v.proj_blk.4.self_attn_norm.weight
F32
F32
[1152]
v.proj_blk.4.self_attn_out.bias
F32
F32
[1152]
v.proj_blk.4.self_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.4.self_attn_q.bias
F32
F32
[1152]
v.proj_blk.4.self_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.4.self_attn_v.bias
F32
F32
[1152]
v.proj_blk.4.self_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.5.cross_attn_k.bias
F32
F32
[1152]
v.proj_blk.5.cross_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.5.cross_attn_norm.bias
F32
F32
[1152]
v.proj_blk.5.cross_attn_norm.weight
F32
F32
[1152]
v.proj_blk.5.cross_attn_out.bias
F32
F32
[1152]
v.proj_blk.5.cross_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.5.cross_attn_q.bias
F32
F32
[1152]
v.proj_blk.5.cross_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.5.cross_attn_v.bias
F32
F32
[1152]
v.proj_blk.5.cross_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.5.ffn_down.bias
F32
F32
[1152]
v.proj_blk.5.ffn_down.weight
F16
F16
[3072, 1152]
v.proj_blk.5.ffn_norm.bias
F32
F32
[1152]
v.proj_blk.5.ffn_norm.weight
F32
F32
[1152]
v.proj_blk.5.ffn_up.bias
F32
F32
[3072]
v.proj_blk.5.ffn_up.weight
F16
F16
[1152, 3072]
v.proj_blk.5.img_pos
F32
F32
[1152, 64, 1]
v.proj_blk.5.linear.bias
F32
F32
[2560]
v.proj_blk.5.linear.weight
F16
F16
[1152, 2560]
v.proj_blk.5.norm.bias
F32
F32
[1152]
v.proj_blk.5.norm.weight
F32
F32
[1152]
v.proj_blk.5.post_norm.bias
F32
F32
[1152]
v.proj_blk.5.post_norm.weight
F32
F32
[1152]
v.proj_blk.5.query
F32
F32
[1152, 16, 1]
v.proj_blk.5.self_attn_k.bias
F32
F32
[1152]
v.proj_blk.5.self_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.5.self_attn_norm.bias
F32
F32
[1152]
v.proj_blk.5.self_attn_norm.weight
F32
F32
[1152]
v.proj_blk.5.self_attn_out.bias
F32
F32
[1152]
v.proj_blk.5.self_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.5.self_attn_q.bias
F32
F32
[1152]
v.proj_blk.5.self_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.5.self_attn_v.bias
F32
F32
[1152]
v.proj_blk.5.self_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.6.cross_attn_k.bias
F32
F32
[1152]
v.proj_blk.6.cross_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.6.cross_attn_norm.bias
F32
F32
[1152]
v.proj_blk.6.cross_attn_norm.weight
F32
F32
[1152]
v.proj_blk.6.cross_attn_out.bias
F32
F32
[1152]
v.proj_blk.6.cross_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.6.cross_attn_q.bias
F32
F32
[1152]
v.proj_blk.6.cross_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.6.cross_attn_v.bias
F32
F32
[1152]
v.proj_blk.6.cross_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.6.ffn_down.bias
F32
F32
[1152]
v.proj_blk.6.ffn_down.weight
F16
F16
[3072, 1152]
v.proj_blk.6.ffn_norm.bias
F32
F32
[1152]
v.proj_blk.6.ffn_norm.weight
F32
F32
[1152]
v.proj_blk.6.ffn_up.bias
F32
F32
[3072]
v.proj_blk.6.ffn_up.weight
F16
F16
[1152, 3072]
v.proj_blk.6.img_pos
F32
F32
[1152, 64, 1]
v.proj_blk.6.linear.bias
F32
F32
[2560]
v.proj_blk.6.linear.weight
F16
F16
[1152, 2560]
v.proj_blk.6.norm.bias
F32
F32
[1152]
v.proj_blk.6.norm.weight
F32
F32
[1152]
v.proj_blk.6.post_norm.bias
F32
F32
[1152]
v.proj_blk.6.post_norm.weight
F32
F32
[1152]
v.proj_blk.6.query
F32
F32
[1152, 16, 1]
v.proj_blk.6.self_attn_k.bias
F32
F32
[1152]
v.proj_blk.6.self_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.6.self_attn_norm.bias
F32
F32
[1152]
v.proj_blk.6.self_attn_norm.weight
F32
F32
[1152]
v.proj_blk.6.self_attn_out.bias
F32
F32
[1152]
v.proj_blk.6.self_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.6.self_attn_q.bias
F32
F32
[1152]
v.proj_blk.6.self_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.6.self_attn_v.bias
F32
F32
[1152]
v.proj_blk.6.self_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.7.cross_attn_k.bias
F32
F32
[1152]
v.proj_blk.7.cross_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.7.cross_attn_norm.bias
F32
F32
[1152]
v.proj_blk.7.cross_attn_norm.weight
F32
F32
[1152]
v.proj_blk.7.cross_attn_out.bias
F32
F32
[1152]
v.proj_blk.7.cross_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.7.cross_attn_q.bias
F32
F32
[1152]
v.proj_blk.7.cross_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.7.cross_attn_v.bias
F32
F32
[1152]
v.proj_blk.7.cross_attn_v.weight
F16
F16
[1152, 1152]
v.proj_blk.7.ffn_down.bias
F32
F32
[1152]
v.proj_blk.7.ffn_down.weight
F16
F16
[3072, 1152]
v.proj_blk.7.ffn_norm.bias
F32
F32
[1152]
v.proj_blk.7.ffn_norm.weight
F32
F32
[1152]
v.proj_blk.7.ffn_up.bias
F32
F32
[3072]
v.proj_blk.7.ffn_up.weight
F16
F16
[1152, 3072]
v.proj_blk.7.img_pos
F32
F32
[1152, 64, 1]
v.proj_blk.7.linear.bias
F32
F32
[2560]
v.proj_blk.7.linear.weight
F16
F16
[1152, 2560]
v.proj_blk.7.norm.bias
F32
F32
[1152]
v.proj_blk.7.norm.weight
F32
F32
[1152]
v.proj_blk.7.post_norm.bias
F32
F32
[1152]
v.proj_blk.7.post_norm.weight
F32
F32
[1152]
v.proj_blk.7.query
F32
F32
[1152, 16, 1]
v.proj_blk.7.self_attn_k.bias
F32
F32
[1152]
v.proj_blk.7.self_attn_k.weight
F16
F16
[1152, 1152]
v.proj_blk.7.self_attn_norm.bias
F32
F32
[1152]
v.proj_blk.7.self_attn_norm.weight
F32
F32
[1152]
v.proj_blk.7.self_attn_out.bias
F32
F32
[1152]
v.proj_blk.7.self_attn_out.weight
F16
F16
[1152, 1152]
v.proj_blk.7.self_attn_q.bias
F32
F32
[1152]
v.proj_blk.7.self_attn_q.weight
F16
F16
[1152, 1152]
v.proj_blk.7.self_attn_v.bias
F32
F32
[1152]
v.proj_blk.7.self_attn_v.weight
F16
F16
[1152, 1152]