Models
GitHub
Discord
Turbo
Sign in
Download
Models
Download
GitHub
Discord
Sign in
pdevine
/
paligemma
:latest
65
Downloads
Updated
5 months ago
Cancel
paligemma:latest
...
/
model
a7f96819fba7 · 5.9GB
Metadata
general.architecture
paligemma
paligemma
general.file_type
F16
F16
paligemma.attention.head_count
8
8
paligemma.attention.head_count_kv
1
1
paligemma.attention.sliding_window
0
0
paligemma.attn_logit_softcapping
0
0
paligemma.block_count
18
18
paligemma.embedding_length
2048
2048
paligemma.feed_forward_length
16384
16384
paligemma.final_logit_softcapping
0
0
paligemma.text.attention.key_length
0
0
paligemma.text.attention.value_length
0
0
tokenizer.ggml.add_bos_token
true
true
tokenizer.ggml.add_eos_token
false
false
tokenizer.ggml.add_padding_token
false
false
tokenizer.ggml.add_unknown_token
false
false
tokenizer.ggml.bos_token_id
2
2
tokenizer.ggml.eos_token_id
1
1
tokenizer.ggml.merges
[ , , , , , ...]
[ , , , , , ...]
tokenizer.ggml.model
llama
llama
tokenizer.ggml.padding_token_id
0
0
tokenizer.ggml.pre
default
default
tokenizer.ggml.scores
[0, 0, 0, 0, 0, ...]
[0, 0, 0, 0, 0, ...]
tokenizer.ggml.token_type
[3, 3, 3, 2, 1, ...]
[3, 3, 3, 2, 1, ...]
tokenizer.ggml.tokens
[<pad>, <eos>, <bos>, <unk>, <mask>, ...]
[<pad>, <eos>, <bos>, <unk>, <mask>, ...]
tokenizer.ggml.unknown_token_id
3
3
Tensor
Name
Type
Shape
token_embd.weight
F16
F16
[2048, 257216]
blk.0
blk.0.attn_k.weight
F16
F16
[2048, 256]
blk.0.attn_norm.weight
F32
F32
[2048]
blk.0.attn_output.weight
F16
F16
[2048, 2048]
blk.0.attn_q.weight
F16
F16
[2048, 2048]
blk.0.attn_v.weight
F16
F16
[2048, 256]
blk.0.ffn_down.weight
F16
F16
[16384, 2048]
blk.0.ffn_gate.weight
F16
F16
[2048, 16384]
blk.0.ffn_up.weight
F16
F16
[2048, 16384]
blk.0.post_attention_norm.weight
F32
F32
[2048]
blk.1
blk.1.attn_k.weight
F16
F16
[2048, 256]
blk.1.attn_norm.weight
F32
F32
[2048]
blk.1.attn_output.weight
F16
F16
[2048, 2048]
blk.1.attn_q.weight
F16
F16
[2048, 2048]
blk.1.attn_v.weight
F16
F16
[2048, 256]
blk.1.ffn_down.weight
F16
F16
[16384, 2048]
blk.1.ffn_gate.weight
F16
F16
[2048, 16384]
blk.1.ffn_up.weight
F16
F16
[2048, 16384]
blk.1.post_attention_norm.weight
F32
F32
[2048]
blk.2
blk.2.attn_k.weight
F16
F16
[2048, 256]
blk.2.attn_norm.weight
F32
F32
[2048]
blk.2.attn_output.weight
F16
F16
[2048, 2048]
blk.2.attn_q.weight
F16
F16
[2048, 2048]
blk.2.attn_v.weight
F16
F16
[2048, 256]
blk.2.ffn_down.weight
F16
F16
[16384, 2048]
blk.2.ffn_gate.weight
F16
F16
[2048, 16384]
blk.2.ffn_up.weight
F16
F16
[2048, 16384]
blk.2.post_attention_norm.weight
F32
F32
[2048]
blk.3
blk.3.attn_k.weight
F16
F16
[2048, 256]
blk.3.attn_norm.weight
F32
F32
[2048]
blk.3.attn_output.weight
F16
F16
[2048, 2048]
blk.3.attn_q.weight
F16
F16
[2048, 2048]
blk.3.attn_v.weight
F16
F16
[2048, 256]
blk.3.ffn_down.weight
F16
F16
[16384, 2048]
blk.3.ffn_gate.weight
F16
F16
[2048, 16384]
blk.3.ffn_up.weight
F16
F16
[2048, 16384]
blk.3.post_attention_norm.weight
F32
F32
[2048]
blk.4
blk.4.attn_k.weight
F16
F16
[2048, 256]
blk.4.attn_norm.weight
F32
F32
[2048]
blk.4.attn_output.weight
F16
F16
[2048, 2048]
blk.4.attn_q.weight
F16
F16
[2048, 2048]
blk.4.attn_v.weight
F16
F16
[2048, 256]
blk.4.ffn_down.weight
F16
F16
[16384, 2048]
blk.4.ffn_gate.weight
F16
F16
[2048, 16384]
blk.4.ffn_up.weight
F16
F16
[2048, 16384]
blk.4.post_attention_norm.weight
F32
F32
[2048]
blk.5
blk.5.attn_k.weight
F16
F16
[2048, 256]
blk.5.attn_norm.weight
F32
F32
[2048]
blk.5.attn_output.weight
F16
F16
[2048, 2048]
blk.5.attn_q.weight
F16
F16
[2048, 2048]
blk.5.attn_v.weight
F16
F16
[2048, 256]
blk.5.ffn_down.weight
F16
F16
[16384, 2048]
blk.5.ffn_gate.weight
F16
F16
[2048, 16384]
blk.5.ffn_up.weight
F16
F16
[2048, 16384]
blk.5.post_attention_norm.weight
F32
F32
[2048]
blk.6
blk.6.attn_k.weight
F16
F16
[2048, 256]
blk.6.attn_norm.weight
F32
F32
[2048]
blk.6.attn_output.weight
F16
F16
[2048, 2048]
blk.6.attn_q.weight
F16
F16
[2048, 2048]
blk.6.attn_v.weight
F16
F16
[2048, 256]
blk.6.ffn_down.weight
F16
F16
[16384, 2048]
blk.6.ffn_gate.weight
F16
F16
[2048, 16384]
blk.6.ffn_up.weight
F16
F16
[2048, 16384]
blk.6.post_attention_norm.weight
F32
F32
[2048]
blk.7
blk.7.attn_k.weight
F16
F16
[2048, 256]
blk.7.attn_norm.weight
F32
F32
[2048]
blk.7.attn_output.weight
F16
F16
[2048, 2048]
blk.7.attn_q.weight
F16
F16
[2048, 2048]
blk.7.attn_v.weight
F16
F16
[2048, 256]
blk.7.ffn_down.weight
F16
F16
[16384, 2048]
blk.7.ffn_gate.weight
F16
F16
[2048, 16384]
blk.7.ffn_up.weight
F16
F16
[2048, 16384]
blk.7.post_attention_norm.weight
F32
F32
[2048]
blk.8
blk.8.attn_k.weight
F16
F16
[2048, 256]
blk.8.attn_norm.weight
F32
F32
[2048]
blk.8.attn_output.weight
F16
F16
[2048, 2048]
blk.8.attn_q.weight
F16
F16
[2048, 2048]
blk.8.attn_v.weight
F16
F16
[2048, 256]
blk.8.ffn_down.weight
F16
F16
[16384, 2048]
blk.8.ffn_gate.weight
F16
F16
[2048, 16384]
blk.8.ffn_up.weight
F16
F16
[2048, 16384]
blk.8.post_attention_norm.weight
F32
F32
[2048]
blk.9
blk.9.attn_k.weight
F16
F16
[2048, 256]
blk.9.attn_norm.weight
F32
F32
[2048]
blk.9.attn_output.weight
F16
F16
[2048, 2048]
blk.9.attn_q.weight
F16
F16
[2048, 2048]
blk.9.attn_v.weight
F16
F16
[2048, 256]
blk.9.ffn_down.weight
F16
F16
[16384, 2048]
blk.9.ffn_gate.weight
F16
F16
[2048, 16384]
blk.9.ffn_up.weight
F16
F16
[2048, 16384]
blk.9.post_attention_norm.weight
F32
F32
[2048]
blk.10
blk.10.attn_k.weight
F16
F16
[2048, 256]
blk.10.attn_norm.weight
F32
F32
[2048]
blk.10.attn_output.weight
F16
F16
[2048, 2048]
blk.10.attn_q.weight
F16
F16
[2048, 2048]
blk.10.attn_v.weight
F16
F16
[2048, 256]
blk.10.ffn_down.weight
F16
F16
[16384, 2048]
blk.10.ffn_gate.weight
F16
F16
[2048, 16384]
blk.10.ffn_up.weight
F16
F16
[2048, 16384]
blk.10.post_attention_norm.weight
F32
F32
[2048]
blk.11
blk.11.attn_k.weight
F16
F16
[2048, 256]
blk.11.attn_norm.weight
F32
F32
[2048]
blk.11.attn_output.weight
F16
F16
[2048, 2048]
blk.11.attn_q.weight
F16
F16
[2048, 2048]
blk.11.attn_v.weight
F16
F16
[2048, 256]
blk.11.ffn_down.weight
F16
F16
[16384, 2048]
blk.11.ffn_gate.weight
F16
F16
[2048, 16384]
blk.11.ffn_up.weight
F16
F16
[2048, 16384]
blk.11.post_attention_norm.weight
F32
F32
[2048]
blk.12
blk.12.attn_k.weight
F16
F16
[2048, 256]
blk.12.attn_norm.weight
F32
F32
[2048]
blk.12.attn_output.weight
F16
F16
[2048, 2048]
blk.12.attn_q.weight
F16
F16
[2048, 2048]
blk.12.attn_v.weight
F16
F16
[2048, 256]
blk.12.ffn_down.weight
F16
F16
[16384, 2048]
blk.12.ffn_gate.weight
F16
F16
[2048, 16384]
blk.12.ffn_up.weight
F16
F16
[2048, 16384]
blk.12.post_attention_norm.weight
F32
F32
[2048]
blk.13
blk.13.attn_k.weight
F16
F16
[2048, 256]
blk.13.attn_norm.weight
F32
F32
[2048]
blk.13.attn_output.weight
F16
F16
[2048, 2048]
blk.13.attn_q.weight
F16
F16
[2048, 2048]
blk.13.attn_v.weight
F16
F16
[2048, 256]
blk.13.ffn_down.weight
F16
F16
[16384, 2048]
blk.13.ffn_gate.weight
F16
F16
[2048, 16384]
blk.13.ffn_up.weight
F16
F16
[2048, 16384]
blk.13.post_attention_norm.weight
F32
F32
[2048]
blk.14
blk.14.attn_k.weight
F16
F16
[2048, 256]
blk.14.attn_norm.weight
F32
F32
[2048]
blk.14.attn_output.weight
F16
F16
[2048, 2048]
blk.14.attn_q.weight
F16
F16
[2048, 2048]
blk.14.attn_v.weight
F16
F16
[2048, 256]
blk.14.ffn_down.weight
F16
F16
[16384, 2048]
blk.14.ffn_gate.weight
F16
F16
[2048, 16384]
blk.14.ffn_up.weight
F16
F16
[2048, 16384]
blk.14.post_attention_norm.weight
F32
F32
[2048]
blk.15
blk.15.attn_k.weight
F16
F16
[2048, 256]
blk.15.attn_norm.weight
F32
F32
[2048]
blk.15.attn_output.weight
F16
F16
[2048, 2048]
blk.15.attn_q.weight
F16
F16
[2048, 2048]
blk.15.attn_v.weight
F16
F16
[2048, 256]
blk.15.ffn_down.weight
F16
F16
[16384, 2048]
blk.15.ffn_gate.weight
F16
F16
[2048, 16384]
blk.15.ffn_up.weight
F16
F16
[2048, 16384]
blk.15.post_attention_norm.weight
F32
F32
[2048]
blk.16
blk.16.attn_k.weight
F16
F16
[2048, 256]
blk.16.attn_norm.weight
F32
F32
[2048]
blk.16.attn_output.weight
F16
F16
[2048, 2048]
blk.16.attn_q.weight
F16
F16
[2048, 2048]
blk.16.attn_v.weight
F16
F16
[2048, 256]
blk.16.ffn_down.weight
F16
F16
[16384, 2048]
blk.16.ffn_gate.weight
F16
F16
[2048, 16384]
blk.16.ffn_up.weight
F16
F16
[2048, 16384]
blk.16.post_attention_norm.weight
F32
F32
[2048]
blk.17
blk.17.attn_k.weight
F16
F16
[2048, 256]
blk.17.attn_norm.weight
F32
F32
[2048]
blk.17.attn_output.weight
F16
F16
[2048, 2048]
blk.17.attn_q.weight
F16
F16
[2048, 2048]
blk.17.attn_v.weight
F16
F16
[2048, 256]
blk.17.ffn_down.weight
F16
F16
[16384, 2048]
blk.17.ffn_gate.weight
F16
F16
[2048, 16384]
blk.17.ffn_up.weight
F16
F16
[2048, 16384]
blk.17.post_attention_norm.weight
F32
F32
[2048]
mm.linear.bias
F32
F32
[2048]
mm.linear.weight
F16
F16
[1152, 2048]
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_output.bias
F32
F32
[1152]
v.blk.0.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.0.layer_norm1.weight
F32
F32
[1152]
v.blk.0.layer_norm2.bias
F32
F32
[1152]
v.blk.0.layer_norm2.weight
F32
F32
[1152]
v.blk.0.mlp.fc1.bias
F32
F32
[4304]
v.blk.0.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.0.mlp.fc2.bias
F32
F32
[1152]
v.blk.0.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.1.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.1.layer_norm1.weight
F32
F32
[1152]
v.blk.1.layer_norm2.bias
F32
F32
[1152]
v.blk.1.layer_norm2.weight
F32
F32
[1152]
v.blk.1.mlp.fc1.bias
F32
F32
[4304]
v.blk.1.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.1.mlp.fc2.bias
F32
F32
[1152]
v.blk.1.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.2.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.2.layer_norm1.weight
F32
F32
[1152]
v.blk.2.layer_norm2.bias
F32
F32
[1152]
v.blk.2.layer_norm2.weight
F32
F32
[1152]
v.blk.2.mlp.fc1.bias
F32
F32
[4304]
v.blk.2.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.2.mlp.fc2.bias
F32
F32
[1152]
v.blk.2.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.3.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.3.layer_norm1.weight
F32
F32
[1152]
v.blk.3.layer_norm2.bias
F32
F32
[1152]
v.blk.3.layer_norm2.weight
F32
F32
[1152]
v.blk.3.mlp.fc1.bias
F32
F32
[4304]
v.blk.3.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.3.mlp.fc2.bias
F32
F32
[1152]
v.blk.3.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.4.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.4.layer_norm1.weight
F32
F32
[1152]
v.blk.4.layer_norm2.bias
F32
F32
[1152]
v.blk.4.layer_norm2.weight
F32
F32
[1152]
v.blk.4.mlp.fc1.bias
F32
F32
[4304]
v.blk.4.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.4.mlp.fc2.bias
F32
F32
[1152]
v.blk.4.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.5.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.5.layer_norm1.weight
F32
F32
[1152]
v.blk.5.layer_norm2.bias
F32
F32
[1152]
v.blk.5.layer_norm2.weight
F32
F32
[1152]
v.blk.5.mlp.fc1.bias
F32
F32
[4304]
v.blk.5.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.5.mlp.fc2.bias
F32
F32
[1152]
v.blk.5.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.6.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.6.layer_norm1.weight
F32
F32
[1152]
v.blk.6.layer_norm2.bias
F32
F32
[1152]
v.blk.6.layer_norm2.weight
F32
F32
[1152]
v.blk.6.mlp.fc1.bias
F32
F32
[4304]
v.blk.6.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.6.mlp.fc2.bias
F32
F32
[1152]
v.blk.6.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.7.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.7.layer_norm1.weight
F32
F32
[1152]
v.blk.7.layer_norm2.bias
F32
F32
[1152]
v.blk.7.layer_norm2.weight
F32
F32
[1152]
v.blk.7.mlp.fc1.bias
F32
F32
[4304]
v.blk.7.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.7.mlp.fc2.bias
F32
F32
[1152]
v.blk.7.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.8.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.8.layer_norm1.weight
F32
F32
[1152]
v.blk.8.layer_norm2.bias
F32
F32
[1152]
v.blk.8.layer_norm2.weight
F32
F32
[1152]
v.blk.8.mlp.fc1.bias
F32
F32
[4304]
v.blk.8.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.8.mlp.fc2.bias
F32
F32
[1152]
v.blk.8.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.9.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.9.layer_norm1.weight
F32
F32
[1152]
v.blk.9.layer_norm2.bias
F32
F32
[1152]
v.blk.9.layer_norm2.weight
F32
F32
[1152]
v.blk.9.mlp.fc1.bias
F32
F32
[4304]
v.blk.9.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.9.mlp.fc2.bias
F32
F32
[1152]
v.blk.9.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.10.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.10.layer_norm1.weight
F32
F32
[1152]
v.blk.10.layer_norm2.bias
F32
F32
[1152]
v.blk.10.layer_norm2.weight
F32
F32
[1152]
v.blk.10.mlp.fc1.bias
F32
F32
[4304]
v.blk.10.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.10.mlp.fc2.bias
F32
F32
[1152]
v.blk.10.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.11.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.11.layer_norm1.weight
F32
F32
[1152]
v.blk.11.layer_norm2.bias
F32
F32
[1152]
v.blk.11.layer_norm2.weight
F32
F32
[1152]
v.blk.11.mlp.fc1.bias
F32
F32
[4304]
v.blk.11.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.11.mlp.fc2.bias
F32
F32
[1152]
v.blk.11.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.12.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.12.layer_norm1.weight
F32
F32
[1152]
v.blk.12.layer_norm2.bias
F32
F32
[1152]
v.blk.12.layer_norm2.weight
F32
F32
[1152]
v.blk.12.mlp.fc1.bias
F32
F32
[4304]
v.blk.12.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.12.mlp.fc2.bias
F32
F32
[1152]
v.blk.12.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.13.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.13.layer_norm1.weight
F32
F32
[1152]
v.blk.13.layer_norm2.bias
F32
F32
[1152]
v.blk.13.layer_norm2.weight
F32
F32
[1152]
v.blk.13.mlp.fc1.bias
F32
F32
[4304]
v.blk.13.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.13.mlp.fc2.bias
F32
F32
[1152]
v.blk.13.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.14.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.14.layer_norm1.weight
F32
F32
[1152]
v.blk.14.layer_norm2.bias
F32
F32
[1152]
v.blk.14.layer_norm2.weight
F32
F32
[1152]
v.blk.14.mlp.fc1.bias
F32
F32
[4304]
v.blk.14.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.14.mlp.fc2.bias
F32
F32
[1152]
v.blk.14.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.15.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.15.layer_norm1.weight
F32
F32
[1152]
v.blk.15.layer_norm2.bias
F32
F32
[1152]
v.blk.15.layer_norm2.weight
F32
F32
[1152]
v.blk.15.mlp.fc1.bias
F32
F32
[4304]
v.blk.15.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.15.mlp.fc2.bias
F32
F32
[1152]
v.blk.15.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.16.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.16.layer_norm1.weight
F32
F32
[1152]
v.blk.16.layer_norm2.bias
F32
F32
[1152]
v.blk.16.layer_norm2.weight
F32
F32
[1152]
v.blk.16.mlp.fc1.bias
F32
F32
[4304]
v.blk.16.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.16.mlp.fc2.bias
F32
F32
[1152]
v.blk.16.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.17.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.17.layer_norm1.weight
F32
F32
[1152]
v.blk.17.layer_norm2.bias
F32
F32
[1152]
v.blk.17.layer_norm2.weight
F32
F32
[1152]
v.blk.17.mlp.fc1.bias
F32
F32
[4304]
v.blk.17.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.17.mlp.fc2.bias
F32
F32
[1152]
v.blk.17.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.18.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.18.layer_norm1.weight
F32
F32
[1152]
v.blk.18.layer_norm2.bias
F32
F32
[1152]
v.blk.18.layer_norm2.weight
F32
F32
[1152]
v.blk.18.mlp.fc1.bias
F32
F32
[4304]
v.blk.18.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.18.mlp.fc2.bias
F32
F32
[1152]
v.blk.18.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.19.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.19.layer_norm1.weight
F32
F32
[1152]
v.blk.19.layer_norm2.bias
F32
F32
[1152]
v.blk.19.layer_norm2.weight
F32
F32
[1152]
v.blk.19.mlp.fc1.bias
F32
F32
[4304]
v.blk.19.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.19.mlp.fc2.bias
F32
F32
[1152]
v.blk.19.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.20.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.20.layer_norm1.weight
F32
F32
[1152]
v.blk.20.layer_norm2.bias
F32
F32
[1152]
v.blk.20.layer_norm2.weight
F32
F32
[1152]
v.blk.20.mlp.fc1.bias
F32
F32
[4304]
v.blk.20.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.20.mlp.fc2.bias
F32
F32
[1152]
v.blk.20.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.21.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.21.layer_norm1.weight
F32
F32
[1152]
v.blk.21.layer_norm2.bias
F32
F32
[1152]
v.blk.21.layer_norm2.weight
F32
F32
[1152]
v.blk.21.mlp.fc1.bias
F32
F32
[4304]
v.blk.21.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.21.mlp.fc2.bias
F32
F32
[1152]
v.blk.21.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.22.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.22.layer_norm1.weight
F32
F32
[1152]
v.blk.22.layer_norm2.bias
F32
F32
[1152]
v.blk.22.layer_norm2.weight
F32
F32
[1152]
v.blk.22.mlp.fc1.bias
F32
F32
[4304]
v.blk.22.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.22.mlp.fc2.bias
F32
F32
[1152]
v.blk.22.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.23.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.23.layer_norm1.weight
F32
F32
[1152]
v.blk.23.layer_norm2.bias
F32
F32
[1152]
v.blk.23.layer_norm2.weight
F32
F32
[1152]
v.blk.23.mlp.fc1.bias
F32
F32
[4304]
v.blk.23.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.23.mlp.fc2.bias
F32
F32
[1152]
v.blk.23.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.24.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.24.layer_norm1.weight
F32
F32
[1152]
v.blk.24.layer_norm2.bias
F32
F32
[1152]
v.blk.24.layer_norm2.weight
F32
F32
[1152]
v.blk.24.mlp.fc1.bias
F32
F32
[4304]
v.blk.24.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.24.mlp.fc2.bias
F32
F32
[1152]
v.blk.24.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.25.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.25.layer_norm1.weight
F32
F32
[1152]
v.blk.25.layer_norm2.bias
F32
F32
[1152]
v.blk.25.layer_norm2.weight
F32
F32
[1152]
v.blk.25.mlp.fc1.bias
F32
F32
[4304]
v.blk.25.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.25.mlp.fc2.bias
F32
F32
[1152]
v.blk.25.mlp.fc2.weight
F16
F16
[4304, 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_output.bias
F32
F32
[1152]
v.blk.26.attn_output.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.layer_norm1.bias
F32
F32
[1152]
v.blk.26.layer_norm1.weight
F32
F32
[1152]
v.blk.26.layer_norm2.bias
F32
F32
[1152]
v.blk.26.layer_norm2.weight
F32
F32
[1152]
v.blk.26.mlp.fc1.bias
F32
F32
[4304]
v.blk.26.mlp.fc1.weight
F16
F16
[1152, 4304]
v.blk.26.mlp.fc2.bias
F32
F32
[1152]
v.blk.26.mlp.fc2.weight
F16
F16
[4304, 1152]
v.embeddings.patch_embedding.bias
F32
F32
[1152]
v.embeddings.patch_embedding.weight
F16
F16
[14, 14, 3, 1152]
v.embeddings.position_embedding.weight
F16
F16
[1152, 256]
v.post_layernorm.bias
F32
F32
[1152]
v.post_layernorm.weight
F32
F32
[1152]
output_norm.weight
F32
F32
[2048]