Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
byczech
/
qwen3.6-mmproj-16G-UD-IQ3_S
:35b
580
Downloads
Updated
3 weeks ago
Qwen 3.6 35B multimodal model with vision support, quantized to IQ3_S and optimized to fit within 16 GB VRAM with Q4_0 KV cache. Supports very large context windows, with the practical maximum depending on the Ollama version, GPU and backend.
Qwen 3.6 35B multimodal model with vision support, quantized to IQ3_S and optimized to fit within 16 GB VRAM with Q4_0 KV cache. Supports very large context windows, with the practical maximum depending on the Ollama version, GPU and backend.
Cancel
vision
tools
thinking
35b
qwen3.6-mmproj-16G-UD-IQ3_S:35b
...
/
projector
30f98c7657a7 · 643MB
Metadata
general.architecture
clip
clip
general.file_type
Q8_0
Q8_0
clip.has_vision_encoder
true
true
clip.projector_type
qwen3vl_merger
qwen3vl_merger
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.feed_forward_length
4304
4304
clip.vision.image_mean
[0.5, 0.5, 0.5]
[0.5, 0.5, 0.5]
clip.vision.image_size
768
768
clip.vision.image_std
[0.5, 0.5, 0.5]
[0.5, 0.5, 0.5]
clip.vision.is_deepstack_layers
[false, false, false, false, false, ...]
[false, false, false, false, false, ...]
clip.vision.patch_size
16
16
clip.vision.projection_dim
2048
2048
clip.vision.spatial_merge_size
2
2
Tensor
Name
Type
Shape
mm.0.bias
F32
F32
[4608]
mm.0.weight
F16
F16
[4608, 4608]
mm.2.bias
F32
F32
[2048]
mm.2.weight
F16
F16
[4608, 2048]
v.blk.0
v.blk.0.attn_out.bias
F32
F32
[1152]
v.blk.0.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.0.attn_qkv.bias
F32
F32
[3456]
v.blk.0.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.1.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.1.attn_qkv.bias
F32
F32
[3456]
v.blk.1.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.2.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.2.attn_qkv.bias
F32
F32
[3456]
v.blk.2.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.3.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.3.attn_qkv.bias
F32
F32
[3456]
v.blk.3.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.4.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.4.attn_qkv.bias
F32
F32
[3456]
v.blk.4.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.5.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.5.attn_qkv.bias
F32
F32
[3456]
v.blk.5.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.6.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.6.attn_qkv.bias
F32
F32
[3456]
v.blk.6.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.7.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.7.attn_qkv.bias
F32
F32
[3456]
v.blk.7.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.8.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.8.attn_qkv.bias
F32
F32
[3456]
v.blk.8.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.9.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.9.attn_qkv.bias
F32
F32
[3456]
v.blk.9.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.10.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.10.attn_qkv.bias
F32
F32
[3456]
v.blk.10.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.11.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.11.attn_qkv.bias
F32
F32
[3456]
v.blk.11.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.12.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.12.attn_qkv.bias
F32
F32
[3456]
v.blk.12.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.13.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.13.attn_qkv.bias
F32
F32
[3456]
v.blk.13.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.14.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.14.attn_qkv.bias
F32
F32
[3456]
v.blk.14.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.15.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.15.attn_qkv.bias
F32
F32
[3456]
v.blk.15.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.16.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.16.attn_qkv.bias
F32
F32
[3456]
v.blk.16.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.17.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.17.attn_qkv.bias
F32
F32
[3456]
v.blk.17.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.18.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.18.attn_qkv.bias
F32
F32
[3456]
v.blk.18.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.19.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.19.attn_qkv.bias
F32
F32
[3456]
v.blk.19.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.20.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.20.attn_qkv.bias
F32
F32
[3456]
v.blk.20.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.21.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.21.attn_qkv.bias
F32
F32
[3456]
v.blk.21.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.22.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.22.attn_qkv.bias
F32
F32
[3456]
v.blk.22.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.23.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.23.attn_qkv.bias
F32
F32
[3456]
v.blk.23.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.24.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.24.attn_qkv.bias
F32
F32
[3456]
v.blk.24.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.25.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.25.attn_qkv.bias
F32
F32
[3456]
v.blk.25.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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_out.bias
F32
F32
[1152]
v.blk.26.attn_out.weight
Q8_0
Q8_0
[1152, 1152]
v.blk.26.attn_qkv.bias
F32
F32
[3456]
v.blk.26.attn_qkv.weight
Q8_0
Q8_0
[1152, 3456]
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
Q8_0
Q8_0
[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.patch_embd.bias
F32
F32
[1152]
v.patch_embd.weight
F32
F32
[16, 16, 3, 1152]
v.patch_embd.weight.1
F32
F32
[16, 16, 3, 1152]
v.position_embd.weight
F32
F32
[1152, 2304]
v.post_ln.bias
F32
F32
[1152]
v.post_ln.weight
F32
F32
[1152]