Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
benhaotang
/
Nanonets-OCR-s
:F16
8,309
Downloads
Updated
11 months ago
OCR model by Nanonets that excels at turning anything into markdown
OCR model by Nanonets that excels at turning anything into markdown
Cancel
vision
Nanonets-OCR-s:F16
...
/
model
627924625740 · 7.5GB
Metadata
general.architecture
qwen25vl
qwen25vl
general.file_type
F16
F16
qwen25vl.attention.head_count
16
16
qwen25vl.attention.head_count_kv
2
2
qwen25vl.attention.layer_norm_rms_epsilon
1e-06
1e-06
qwen25vl.block_count
36
36
qwen25vl.context_length
128000
128000
qwen25vl.embedding_length
2048
2048
qwen25vl.feed_forward_length
11008
11008
qwen25vl.rope.freq_base
1e+06
1e+06
qwen25vl.rope.mrope_section
[16, 24, 24]
[16, 24, 24]
qwen25vl.vision.attention.head_count
16
16
qwen25vl.vision.attention.layer_norm_epsilon
1e-06
1e-06
qwen25vl.vision.block_count
32
32
qwen25vl.vision.embedding_length
1280
1280
qwen25vl.vision.fullatt_block_indexes
[7, 15, 23, 31]
[7, 15, 23, 31]
qwen25vl.vision.num_channels
3
3
qwen25vl.vision.patch_size
14
14
qwen25vl.vision.rope.freq_base
10000
10000
qwen25vl.vision.spatial_merge_size
2
2
qwen25vl.vision.spatial_patch_size
14
14
qwen25vl.vision.temporal_patch_size
2
2
qwen25vl.vision.window_size
112
112
tokenizer.ggml.add_eos_token
false
false
tokenizer.ggml.add_padding_token
false
false
tokenizer.ggml.eos_token_id
151645
151645
tokenizer.ggml.eos_token_ids
[151645, 151643]
[151645, 151643]
tokenizer.ggml.merges
[Ġ Ġ, ĠĠ ĠĠ, i n, Ġ t, ĠĠĠĠ ĠĠĠĠ, ...]
[Ġ Ġ, ĠĠ ĠĠ, i n, Ġ t, ĠĠĠĠ ĠĠĠĠ, ...]
tokenizer.ggml.model
gpt2
gpt2
tokenizer.ggml.padding_token_id
151643
151643
tokenizer.ggml.pre
qwen2
qwen2
tokenizer.ggml.scores
[0, 1, 2, 3, 4, ...]
[0, 1, 2, 3, 4, ...]
tokenizer.ggml.token_type
[1, 1, 1, 1, 1, ...]
[1, 1, 1, 1, 1, ...]
tokenizer.ggml.tokens
[!, ", #, $, %, ...]
[!, ", #, $, %, ...]
Tensor
Name
Type
Shape
token_embd.weight
F16
F16
[2048, 151936]
blk.0
blk.0.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.0.attn_q.weight
F16
F16
[2048, 2048]
blk.0.attn_v.bias
F32
F32
[256]
blk.0.attn_v.weight
F16
F16
[2048, 256]
blk.0.ffn_down.weight
F16
F16
[11008, 2048]
blk.0.ffn_gate.weight
F16
F16
[2048, 11008]
blk.0.ffn_norm.weight
F32
F32
[2048]
blk.0.ffn_up.weight
F16
F16
[2048, 11008]
blk.1
blk.1.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.1.attn_q.weight
F16
F16
[2048, 2048]
blk.1.attn_v.bias
F32
F32
[256]
blk.1.attn_v.weight
F16
F16
[2048, 256]
blk.1.ffn_down.weight
F16
F16
[11008, 2048]
blk.1.ffn_gate.weight
F16
F16
[2048, 11008]
blk.1.ffn_norm.weight
F32
F32
[2048]
blk.1.ffn_up.weight
F16
F16
[2048, 11008]
blk.2
blk.2.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.2.attn_q.weight
F16
F16
[2048, 2048]
blk.2.attn_v.bias
F32
F32
[256]
blk.2.attn_v.weight
F16
F16
[2048, 256]
blk.2.ffn_down.weight
F16
F16
[11008, 2048]
blk.2.ffn_gate.weight
F16
F16
[2048, 11008]
blk.2.ffn_norm.weight
F32
F32
[2048]
blk.2.ffn_up.weight
F16
F16
[2048, 11008]
blk.3
blk.3.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.3.attn_q.weight
F16
F16
[2048, 2048]
blk.3.attn_v.bias
F32
F32
[256]
blk.3.attn_v.weight
F16
F16
[2048, 256]
blk.3.ffn_down.weight
F16
F16
[11008, 2048]
blk.3.ffn_gate.weight
F16
F16
[2048, 11008]
blk.3.ffn_norm.weight
F32
F32
[2048]
blk.3.ffn_up.weight
F16
F16
[2048, 11008]
blk.4
blk.4.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.4.attn_q.weight
F16
F16
[2048, 2048]
blk.4.attn_v.bias
F32
F32
[256]
blk.4.attn_v.weight
F16
F16
[2048, 256]
blk.4.ffn_down.weight
F16
F16
[11008, 2048]
blk.4.ffn_gate.weight
F16
F16
[2048, 11008]
blk.4.ffn_norm.weight
F32
F32
[2048]
blk.4.ffn_up.weight
F16
F16
[2048, 11008]
blk.5
blk.5.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.5.attn_q.weight
F16
F16
[2048, 2048]
blk.5.attn_v.bias
F32
F32
[256]
blk.5.attn_v.weight
F16
F16
[2048, 256]
blk.5.ffn_down.weight
F16
F16
[11008, 2048]
blk.5.ffn_gate.weight
F16
F16
[2048, 11008]
blk.5.ffn_norm.weight
F32
F32
[2048]
blk.5.ffn_up.weight
F16
F16
[2048, 11008]
blk.6
blk.6.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.6.attn_q.weight
F16
F16
[2048, 2048]
blk.6.attn_v.bias
F32
F32
[256]
blk.6.attn_v.weight
F16
F16
[2048, 256]
blk.6.ffn_down.weight
F16
F16
[11008, 2048]
blk.6.ffn_gate.weight
F16
F16
[2048, 11008]
blk.6.ffn_norm.weight
F32
F32
[2048]
blk.6.ffn_up.weight
F16
F16
[2048, 11008]
blk.7
blk.7.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.7.attn_q.weight
F16
F16
[2048, 2048]
blk.7.attn_v.bias
F32
F32
[256]
blk.7.attn_v.weight
F16
F16
[2048, 256]
blk.7.ffn_down.weight
F16
F16
[11008, 2048]
blk.7.ffn_gate.weight
F16
F16
[2048, 11008]
blk.7.ffn_norm.weight
F32
F32
[2048]
blk.7.ffn_up.weight
F16
F16
[2048, 11008]
blk.8
blk.8.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.8.attn_q.weight
F16
F16
[2048, 2048]
blk.8.attn_v.bias
F32
F32
[256]
blk.8.attn_v.weight
F16
F16
[2048, 256]
blk.8.ffn_down.weight
F16
F16
[11008, 2048]
blk.8.ffn_gate.weight
F16
F16
[2048, 11008]
blk.8.ffn_norm.weight
F32
F32
[2048]
blk.8.ffn_up.weight
F16
F16
[2048, 11008]
blk.9
blk.9.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.9.attn_q.weight
F16
F16
[2048, 2048]
blk.9.attn_v.bias
F32
F32
[256]
blk.9.attn_v.weight
F16
F16
[2048, 256]
blk.9.ffn_down.weight
F16
F16
[11008, 2048]
blk.9.ffn_gate.weight
F16
F16
[2048, 11008]
blk.9.ffn_norm.weight
F32
F32
[2048]
blk.9.ffn_up.weight
F16
F16
[2048, 11008]
blk.10
blk.10.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.10.attn_q.weight
F16
F16
[2048, 2048]
blk.10.attn_v.bias
F32
F32
[256]
blk.10.attn_v.weight
F16
F16
[2048, 256]
blk.10.ffn_down.weight
F16
F16
[11008, 2048]
blk.10.ffn_gate.weight
F16
F16
[2048, 11008]
blk.10.ffn_norm.weight
F32
F32
[2048]
blk.10.ffn_up.weight
F16
F16
[2048, 11008]
blk.11
blk.11.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.11.attn_q.weight
F16
F16
[2048, 2048]
blk.11.attn_v.bias
F32
F32
[256]
blk.11.attn_v.weight
F16
F16
[2048, 256]
blk.11.ffn_down.weight
F16
F16
[11008, 2048]
blk.11.ffn_gate.weight
F16
F16
[2048, 11008]
blk.11.ffn_norm.weight
F32
F32
[2048]
blk.11.ffn_up.weight
F16
F16
[2048, 11008]
blk.12
blk.12.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.12.attn_q.weight
F16
F16
[2048, 2048]
blk.12.attn_v.bias
F32
F32
[256]
blk.12.attn_v.weight
F16
F16
[2048, 256]
blk.12.ffn_down.weight
F16
F16
[11008, 2048]
blk.12.ffn_gate.weight
F16
F16
[2048, 11008]
blk.12.ffn_norm.weight
F32
F32
[2048]
blk.12.ffn_up.weight
F16
F16
[2048, 11008]
blk.13
blk.13.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.13.attn_q.weight
F16
F16
[2048, 2048]
blk.13.attn_v.bias
F32
F32
[256]
blk.13.attn_v.weight
F16
F16
[2048, 256]
blk.13.ffn_down.weight
F16
F16
[11008, 2048]
blk.13.ffn_gate.weight
F16
F16
[2048, 11008]
blk.13.ffn_norm.weight
F32
F32
[2048]
blk.13.ffn_up.weight
F16
F16
[2048, 11008]
blk.14
blk.14.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.14.attn_q.weight
F16
F16
[2048, 2048]
blk.14.attn_v.bias
F32
F32
[256]
blk.14.attn_v.weight
F16
F16
[2048, 256]
blk.14.ffn_down.weight
F16
F16
[11008, 2048]
blk.14.ffn_gate.weight
F16
F16
[2048, 11008]
blk.14.ffn_norm.weight
F32
F32
[2048]
blk.14.ffn_up.weight
F16
F16
[2048, 11008]
blk.15
blk.15.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.15.attn_q.weight
F16
F16
[2048, 2048]
blk.15.attn_v.bias
F32
F32
[256]
blk.15.attn_v.weight
F16
F16
[2048, 256]
blk.15.ffn_down.weight
F16
F16
[11008, 2048]
blk.15.ffn_gate.weight
F16
F16
[2048, 11008]
blk.15.ffn_norm.weight
F32
F32
[2048]
blk.15.ffn_up.weight
F16
F16
[2048, 11008]
blk.16
blk.16.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.16.attn_q.weight
F16
F16
[2048, 2048]
blk.16.attn_v.bias
F32
F32
[256]
blk.16.attn_v.weight
F16
F16
[2048, 256]
blk.16.ffn_down.weight
F16
F16
[11008, 2048]
blk.16.ffn_gate.weight
F16
F16
[2048, 11008]
blk.16.ffn_norm.weight
F32
F32
[2048]
blk.16.ffn_up.weight
F16
F16
[2048, 11008]
blk.17
blk.17.attn_k.bias
F32
F32
[256]
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.bias
F32
F32
[2048]
blk.17.attn_q.weight
F16
F16
[2048, 2048]
blk.17.attn_v.bias
F32
F32
[256]
blk.17.attn_v.weight
F16
F16
[2048, 256]
blk.17.ffn_down.weight
F16
F16
[11008, 2048]
blk.17.ffn_gate.weight
F16
F16
[2048, 11008]
blk.17.ffn_norm.weight
F32
F32
[2048]
blk.17.ffn_up.weight
F16
F16
[2048, 11008]
blk.18
blk.18.attn_k.bias
F32
F32
[256]
blk.18.attn_k.weight
F16
F16
[2048, 256]
blk.18.attn_norm.weight
F32
F32
[2048]
blk.18.attn_output.weight
F16
F16
[2048, 2048]
blk.18.attn_q.bias
F32
F32
[2048]
blk.18.attn_q.weight
F16
F16
[2048, 2048]
blk.18.attn_v.bias
F32
F32
[256]
blk.18.attn_v.weight
F16
F16
[2048, 256]
blk.18.ffn_down.weight
F16
F16
[11008, 2048]
blk.18.ffn_gate.weight
F16
F16
[2048, 11008]
blk.18.ffn_norm.weight
F32
F32
[2048]
blk.18.ffn_up.weight
F16
F16
[2048, 11008]
blk.19
blk.19.attn_k.bias
F32
F32
[256]
blk.19.attn_k.weight
F16
F16
[2048, 256]
blk.19.attn_norm.weight
F32
F32
[2048]
blk.19.attn_output.weight
F16
F16
[2048, 2048]
blk.19.attn_q.bias
F32
F32
[2048]
blk.19.attn_q.weight
F16
F16
[2048, 2048]
blk.19.attn_v.bias
F32
F32
[256]
blk.19.attn_v.weight
F16
F16
[2048, 256]
blk.19.ffn_down.weight
F16
F16
[11008, 2048]
blk.19.ffn_gate.weight
F16
F16
[2048, 11008]
blk.19.ffn_norm.weight
F32
F32
[2048]
blk.19.ffn_up.weight
F16
F16
[2048, 11008]
blk.20
blk.20.attn_k.bias
F32
F32
[256]
blk.20.attn_k.weight
F16
F16
[2048, 256]
blk.20.attn_norm.weight
F32
F32
[2048]
blk.20.attn_output.weight
F16
F16
[2048, 2048]
blk.20.attn_q.bias
F32
F32
[2048]
blk.20.attn_q.weight
F16
F16
[2048, 2048]
blk.20.attn_v.bias
F32
F32
[256]
blk.20.attn_v.weight
F16
F16
[2048, 256]
blk.20.ffn_down.weight
F16
F16
[11008, 2048]
blk.20.ffn_gate.weight
F16
F16
[2048, 11008]
blk.20.ffn_norm.weight
F32
F32
[2048]
blk.20.ffn_up.weight
F16
F16
[2048, 11008]
blk.21
blk.21.attn_k.bias
F32
F32
[256]
blk.21.attn_k.weight
F16
F16
[2048, 256]
blk.21.attn_norm.weight
F32
F32
[2048]
blk.21.attn_output.weight
F16
F16
[2048, 2048]
blk.21.attn_q.bias
F32
F32
[2048]
blk.21.attn_q.weight
F16
F16
[2048, 2048]
blk.21.attn_v.bias
F32
F32
[256]
blk.21.attn_v.weight
F16
F16
[2048, 256]
blk.21.ffn_down.weight
F16
F16
[11008, 2048]
blk.21.ffn_gate.weight
F16
F16
[2048, 11008]
blk.21.ffn_norm.weight
F32
F32
[2048]
blk.21.ffn_up.weight
F16
F16
[2048, 11008]
blk.22
blk.22.attn_k.bias
F32
F32
[256]
blk.22.attn_k.weight
F16
F16
[2048, 256]
blk.22.attn_norm.weight
F32
F32
[2048]
blk.22.attn_output.weight
F16
F16
[2048, 2048]
blk.22.attn_q.bias
F32
F32
[2048]
blk.22.attn_q.weight
F16
F16
[2048, 2048]
blk.22.attn_v.bias
F32
F32
[256]
blk.22.attn_v.weight
F16
F16
[2048, 256]
blk.22.ffn_down.weight
F16
F16
[11008, 2048]
blk.22.ffn_gate.weight
F16
F16
[2048, 11008]
blk.22.ffn_norm.weight
F32
F32
[2048]
blk.22.ffn_up.weight
F16
F16
[2048, 11008]
blk.23
blk.23.attn_k.bias
F32
F32
[256]
blk.23.attn_k.weight
F16
F16
[2048, 256]
blk.23.attn_norm.weight
F32
F32
[2048]
blk.23.attn_output.weight
F16
F16
[2048, 2048]
blk.23.attn_q.bias
F32
F32
[2048]
blk.23.attn_q.weight
F16
F16
[2048, 2048]
blk.23.attn_v.bias
F32
F32
[256]
blk.23.attn_v.weight
F16
F16
[2048, 256]
blk.23.ffn_down.weight
F16
F16
[11008, 2048]
blk.23.ffn_gate.weight
F16
F16
[2048, 11008]
blk.23.ffn_norm.weight
F32
F32
[2048]
blk.23.ffn_up.weight
F16
F16
[2048, 11008]
blk.24
blk.24.attn_k.bias
F32
F32
[256]
blk.24.attn_k.weight
F16
F16
[2048, 256]
blk.24.attn_norm.weight
F32
F32
[2048]
blk.24.attn_output.weight
F16
F16
[2048, 2048]
blk.24.attn_q.bias
F32
F32
[2048]
blk.24.attn_q.weight
F16
F16
[2048, 2048]
blk.24.attn_v.bias
F32
F32
[256]
blk.24.attn_v.weight
F16
F16
[2048, 256]
blk.24.ffn_down.weight
F16
F16
[11008, 2048]
blk.24.ffn_gate.weight
F16
F16
[2048, 11008]
blk.24.ffn_norm.weight
F32
F32
[2048]
blk.24.ffn_up.weight
F16
F16
[2048, 11008]
blk.25
blk.25.attn_k.bias
F32
F32
[256]
blk.25.attn_k.weight
F16
F16
[2048, 256]
blk.25.attn_norm.weight
F32
F32
[2048]
blk.25.attn_output.weight
F16
F16
[2048, 2048]
blk.25.attn_q.bias
F32
F32
[2048]
blk.25.attn_q.weight
F16
F16
[2048, 2048]
blk.25.attn_v.bias
F32
F32
[256]
blk.25.attn_v.weight
F16
F16
[2048, 256]
blk.25.ffn_down.weight
F16
F16
[11008, 2048]
blk.25.ffn_gate.weight
F16
F16
[2048, 11008]
blk.25.ffn_norm.weight
F32
F32
[2048]
blk.25.ffn_up.weight
F16
F16
[2048, 11008]
blk.26
blk.26.attn_k.bias
F32
F32
[256]
blk.26.attn_k.weight
F16
F16
[2048, 256]
blk.26.attn_norm.weight
F32
F32
[2048]
blk.26.attn_output.weight
F16
F16
[2048, 2048]
blk.26.attn_q.bias
F32
F32
[2048]
blk.26.attn_q.weight
F16
F16
[2048, 2048]
blk.26.attn_v.bias
F32
F32
[256]
blk.26.attn_v.weight
F16
F16
[2048, 256]
blk.26.ffn_down.weight
F16
F16
[11008, 2048]
blk.26.ffn_gate.weight
F16
F16
[2048, 11008]
blk.26.ffn_norm.weight
F32
F32
[2048]
blk.26.ffn_up.weight
F16
F16
[2048, 11008]
blk.27
blk.27.attn_k.bias
F32
F32
[256]
blk.27.attn_k.weight
F16
F16
[2048, 256]
blk.27.attn_norm.weight
F32
F32
[2048]
blk.27.attn_output.weight
F16
F16
[2048, 2048]
blk.27.attn_q.bias
F32
F32
[2048]
blk.27.attn_q.weight
F16
F16
[2048, 2048]
blk.27.attn_v.bias
F32
F32
[256]
blk.27.attn_v.weight
F16
F16
[2048, 256]
blk.27.ffn_down.weight
F16
F16
[11008, 2048]
blk.27.ffn_gate.weight
F16
F16
[2048, 11008]
blk.27.ffn_norm.weight
F32
F32
[2048]
blk.27.ffn_up.weight
F16
F16
[2048, 11008]
blk.28
blk.28.attn_k.bias
F32
F32
[256]
blk.28.attn_k.weight
F16
F16
[2048, 256]
blk.28.attn_norm.weight
F32
F32
[2048]
blk.28.attn_output.weight
F16
F16
[2048, 2048]
blk.28.attn_q.bias
F32
F32
[2048]
blk.28.attn_q.weight
F16
F16
[2048, 2048]
blk.28.attn_v.bias
F32
F32
[256]
blk.28.attn_v.weight
F16
F16
[2048, 256]
blk.28.ffn_down.weight
F16
F16
[11008, 2048]
blk.28.ffn_gate.weight
F16
F16
[2048, 11008]
blk.28.ffn_norm.weight
F32
F32
[2048]
blk.28.ffn_up.weight
F16
F16
[2048, 11008]
blk.29
blk.29.attn_k.bias
F32
F32
[256]
blk.29.attn_k.weight
F16
F16
[2048, 256]
blk.29.attn_norm.weight
F32
F32
[2048]
blk.29.attn_output.weight
F16
F16
[2048, 2048]
blk.29.attn_q.bias
F32
F32
[2048]
blk.29.attn_q.weight
F16
F16
[2048, 2048]
blk.29.attn_v.bias
F32
F32
[256]
blk.29.attn_v.weight
F16
F16
[2048, 256]
blk.29.ffn_down.weight
F16
F16
[11008, 2048]
blk.29.ffn_gate.weight
F16
F16
[2048, 11008]
blk.29.ffn_norm.weight
F32
F32
[2048]
blk.29.ffn_up.weight
F16
F16
[2048, 11008]
blk.30
blk.30.attn_k.bias
F32
F32
[256]
blk.30.attn_k.weight
F16
F16
[2048, 256]
blk.30.attn_norm.weight
F32
F32
[2048]
blk.30.attn_output.weight
F16
F16
[2048, 2048]
blk.30.attn_q.bias
F32
F32
[2048]
blk.30.attn_q.weight
F16
F16
[2048, 2048]
blk.30.attn_v.bias
F32
F32
[256]
blk.30.attn_v.weight
F16
F16
[2048, 256]
blk.30.ffn_down.weight
F16
F16
[11008, 2048]
blk.30.ffn_gate.weight
F16
F16
[2048, 11008]
blk.30.ffn_norm.weight
F32
F32
[2048]
blk.30.ffn_up.weight
F16
F16
[2048, 11008]
blk.31
blk.31.attn_k.bias
F32
F32
[256]
blk.31.attn_k.weight
F16
F16
[2048, 256]
blk.31.attn_norm.weight
F32
F32
[2048]
blk.31.attn_output.weight
F16
F16
[2048, 2048]
blk.31.attn_q.bias
F32
F32
[2048]
blk.31.attn_q.weight
F16
F16
[2048, 2048]
blk.31.attn_v.bias
F32
F32
[256]
blk.31.attn_v.weight
F16
F16
[2048, 256]
blk.31.ffn_down.weight
F16
F16
[11008, 2048]
blk.31.ffn_gate.weight
F16
F16
[2048, 11008]
blk.31.ffn_norm.weight
F32
F32
[2048]
blk.31.ffn_up.weight
F16
F16
[2048, 11008]
blk.32
blk.32.attn_k.bias
F32
F32
[256]
blk.32.attn_k.weight
F16
F16
[2048, 256]
blk.32.attn_norm.weight
F32
F32
[2048]
blk.32.attn_output.weight
F16
F16
[2048, 2048]
blk.32.attn_q.bias
F32
F32
[2048]
blk.32.attn_q.weight
F16
F16
[2048, 2048]
blk.32.attn_v.bias
F32
F32
[256]
blk.32.attn_v.weight
F16
F16
[2048, 256]
blk.32.ffn_down.weight
F16
F16
[11008, 2048]
blk.32.ffn_gate.weight
F16
F16
[2048, 11008]
blk.32.ffn_norm.weight
F32
F32
[2048]
blk.32.ffn_up.weight
F16
F16
[2048, 11008]
blk.33
blk.33.attn_k.bias
F32
F32
[256]
blk.33.attn_k.weight
F16
F16
[2048, 256]
blk.33.attn_norm.weight
F32
F32
[2048]
blk.33.attn_output.weight
F16
F16
[2048, 2048]
blk.33.attn_q.bias
F32
F32
[2048]
blk.33.attn_q.weight
F16
F16
[2048, 2048]
blk.33.attn_v.bias
F32
F32
[256]
blk.33.attn_v.weight
F16
F16
[2048, 256]
blk.33.ffn_down.weight
F16
F16
[11008, 2048]
blk.33.ffn_gate.weight
F16
F16
[2048, 11008]
blk.33.ffn_norm.weight
F32
F32
[2048]
blk.33.ffn_up.weight
F16
F16
[2048, 11008]
blk.34
blk.34.attn_k.bias
F32
F32
[256]
blk.34.attn_k.weight
F16
F16
[2048, 256]
blk.34.attn_norm.weight
F32
F32
[2048]
blk.34.attn_output.weight
F16
F16
[2048, 2048]
blk.34.attn_q.bias
F32
F32
[2048]
blk.34.attn_q.weight
F16
F16
[2048, 2048]
blk.34.attn_v.bias
F32
F32
[256]
blk.34.attn_v.weight
F16
F16
[2048, 256]
blk.34.ffn_down.weight
F16
F16
[11008, 2048]
blk.34.ffn_gate.weight
F16
F16
[2048, 11008]
blk.34.ffn_norm.weight
F32
F32
[2048]
blk.34.ffn_up.weight
F16
F16
[2048, 11008]
blk.35
blk.35.attn_k.bias
F32
F32
[256]
blk.35.attn_k.weight
F16
F16
[2048, 256]
blk.35.attn_norm.weight
F32
F32
[2048]
blk.35.attn_output.weight
F16
F16
[2048, 2048]
blk.35.attn_q.bias
F32
F32
[2048]
blk.35.attn_q.weight
F16
F16
[2048, 2048]
blk.35.attn_v.bias
F32
F32
[256]
blk.35.attn_v.weight
F16
F16
[2048, 256]
blk.35.ffn_down.weight
F16
F16
[11008, 2048]
blk.35.ffn_gate.weight
F16
F16
[2048, 11008]
blk.35.ffn_norm.weight
F32
F32
[2048]
blk.35.ffn_up.weight
F16
F16
[2048, 11008]
v.blk.0
v.blk.0.attn_k.bias
F32
F32
[1280]
v.blk.0.attn_k.weight
F16
F16
[1280, 1280]
v.blk.0.attn_out.bias
F32
F32
[1280]
v.blk.0.attn_out.weight
F16
F16
[1280, 1280]
v.blk.0.attn_q.bias
F32
F32
[1280]
v.blk.0.attn_q.weight
F16
F16
[1280, 1280]
v.blk.0.attn_v.bias
F32
F32
[1280]
v.blk.0.attn_v.weight
F16
F16
[1280, 1280]
v.blk.0.ffn_down.bias
F32
F32
[1280]
v.blk.0.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.0.ffn_gate.bias
F32
F32
[3420]
v.blk.0.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.0.ffn_up.bias
F32
F32
[3420]
v.blk.0.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.0.ln1.weight
F32
F32
[1280]
v.blk.0.ln2.weight
F32
F32
[1280]
v.blk.1
v.blk.1.attn_k.bias
F32
F32
[1280]
v.blk.1.attn_k.weight
F16
F16
[1280, 1280]
v.blk.1.attn_out.bias
F32
F32
[1280]
v.blk.1.attn_out.weight
F16
F16
[1280, 1280]
v.blk.1.attn_q.bias
F32
F32
[1280]
v.blk.1.attn_q.weight
F16
F16
[1280, 1280]
v.blk.1.attn_v.bias
F32
F32
[1280]
v.blk.1.attn_v.weight
F16
F16
[1280, 1280]
v.blk.1.ffn_down.bias
F32
F32
[1280]
v.blk.1.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.1.ffn_gate.bias
F32
F32
[3420]
v.blk.1.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.1.ffn_up.bias
F32
F32
[3420]
v.blk.1.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.1.ln1.weight
F32
F32
[1280]
v.blk.1.ln2.weight
F32
F32
[1280]
v.blk.2
v.blk.2.attn_k.bias
F32
F32
[1280]
v.blk.2.attn_k.weight
F16
F16
[1280, 1280]
v.blk.2.attn_out.bias
F32
F32
[1280]
v.blk.2.attn_out.weight
F16
F16
[1280, 1280]
v.blk.2.attn_q.bias
F32
F32
[1280]
v.blk.2.attn_q.weight
F16
F16
[1280, 1280]
v.blk.2.attn_v.bias
F32
F32
[1280]
v.blk.2.attn_v.weight
F16
F16
[1280, 1280]
v.blk.2.ffn_down.bias
F32
F32
[1280]
v.blk.2.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.2.ffn_gate.bias
F32
F32
[3420]
v.blk.2.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.2.ffn_up.bias
F32
F32
[3420]
v.blk.2.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.2.ln1.weight
F32
F32
[1280]
v.blk.2.ln2.weight
F32
F32
[1280]
v.blk.3
v.blk.3.attn_k.bias
F32
F32
[1280]
v.blk.3.attn_k.weight
F16
F16
[1280, 1280]
v.blk.3.attn_out.bias
F32
F32
[1280]
v.blk.3.attn_out.weight
F16
F16
[1280, 1280]
v.blk.3.attn_q.bias
F32
F32
[1280]
v.blk.3.attn_q.weight
F16
F16
[1280, 1280]
v.blk.3.attn_v.bias
F32
F32
[1280]
v.blk.3.attn_v.weight
F16
F16
[1280, 1280]
v.blk.3.ffn_down.bias
F32
F32
[1280]
v.blk.3.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.3.ffn_gate.bias
F32
F32
[3420]
v.blk.3.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.3.ffn_up.bias
F32
F32
[3420]
v.blk.3.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.3.ln1.weight
F32
F32
[1280]
v.blk.3.ln2.weight
F32
F32
[1280]
v.blk.4
v.blk.4.attn_k.bias
F32
F32
[1280]
v.blk.4.attn_k.weight
F16
F16
[1280, 1280]
v.blk.4.attn_out.bias
F32
F32
[1280]
v.blk.4.attn_out.weight
F16
F16
[1280, 1280]
v.blk.4.attn_q.bias
F32
F32
[1280]
v.blk.4.attn_q.weight
F16
F16
[1280, 1280]
v.blk.4.attn_v.bias
F32
F32
[1280]
v.blk.4.attn_v.weight
F16
F16
[1280, 1280]
v.blk.4.ffn_down.bias
F32
F32
[1280]
v.blk.4.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.4.ffn_gate.bias
F32
F32
[3420]
v.blk.4.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.4.ffn_up.bias
F32
F32
[3420]
v.blk.4.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.4.ln1.weight
F32
F32
[1280]
v.blk.4.ln2.weight
F32
F32
[1280]
v.blk.5
v.blk.5.attn_k.bias
F32
F32
[1280]
v.blk.5.attn_k.weight
F16
F16
[1280, 1280]
v.blk.5.attn_out.bias
F32
F32
[1280]
v.blk.5.attn_out.weight
F16
F16
[1280, 1280]
v.blk.5.attn_q.bias
F32
F32
[1280]
v.blk.5.attn_q.weight
F16
F16
[1280, 1280]
v.blk.5.attn_v.bias
F32
F32
[1280]
v.blk.5.attn_v.weight
F16
F16
[1280, 1280]
v.blk.5.ffn_down.bias
F32
F32
[1280]
v.blk.5.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.5.ffn_gate.bias
F32
F32
[3420]
v.blk.5.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.5.ffn_up.bias
F32
F32
[3420]
v.blk.5.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.5.ln1.weight
F32
F32
[1280]
v.blk.5.ln2.weight
F32
F32
[1280]
v.blk.6
v.blk.6.attn_k.bias
F32
F32
[1280]
v.blk.6.attn_k.weight
F16
F16
[1280, 1280]
v.blk.6.attn_out.bias
F32
F32
[1280]
v.blk.6.attn_out.weight
F16
F16
[1280, 1280]
v.blk.6.attn_q.bias
F32
F32
[1280]
v.blk.6.attn_q.weight
F16
F16
[1280, 1280]
v.blk.6.attn_v.bias
F32
F32
[1280]
v.blk.6.attn_v.weight
F16
F16
[1280, 1280]
v.blk.6.ffn_down.bias
F32
F32
[1280]
v.blk.6.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.6.ffn_gate.bias
F32
F32
[3420]
v.blk.6.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.6.ffn_up.bias
F32
F32
[3420]
v.blk.6.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.6.ln1.weight
F32
F32
[1280]
v.blk.6.ln2.weight
F32
F32
[1280]
v.blk.7
v.blk.7.attn_k.bias
F32
F32
[1280]
v.blk.7.attn_k.weight
F16
F16
[1280, 1280]
v.blk.7.attn_out.bias
F32
F32
[1280]
v.blk.7.attn_out.weight
F16
F16
[1280, 1280]
v.blk.7.attn_q.bias
F32
F32
[1280]
v.blk.7.attn_q.weight
F16
F16
[1280, 1280]
v.blk.7.attn_v.bias
F32
F32
[1280]
v.blk.7.attn_v.weight
F16
F16
[1280, 1280]
v.blk.7.ffn_down.bias
F32
F32
[1280]
v.blk.7.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.7.ffn_gate.bias
F32
F32
[3420]
v.blk.7.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.7.ffn_up.bias
F32
F32
[3420]
v.blk.7.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.7.ln1.weight
F32
F32
[1280]
v.blk.7.ln2.weight
F32
F32
[1280]
v.blk.8
v.blk.8.attn_k.bias
F32
F32
[1280]
v.blk.8.attn_k.weight
F16
F16
[1280, 1280]
v.blk.8.attn_out.bias
F32
F32
[1280]
v.blk.8.attn_out.weight
F16
F16
[1280, 1280]
v.blk.8.attn_q.bias
F32
F32
[1280]
v.blk.8.attn_q.weight
F16
F16
[1280, 1280]
v.blk.8.attn_v.bias
F32
F32
[1280]
v.blk.8.attn_v.weight
F16
F16
[1280, 1280]
v.blk.8.ffn_down.bias
F32
F32
[1280]
v.blk.8.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.8.ffn_gate.bias
F32
F32
[3420]
v.blk.8.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.8.ffn_up.bias
F32
F32
[3420]
v.blk.8.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.8.ln1.weight
F32
F32
[1280]
v.blk.8.ln2.weight
F32
F32
[1280]
v.blk.9
v.blk.9.attn_k.bias
F32
F32
[1280]
v.blk.9.attn_k.weight
F16
F16
[1280, 1280]
v.blk.9.attn_out.bias
F32
F32
[1280]
v.blk.9.attn_out.weight
F16
F16
[1280, 1280]
v.blk.9.attn_q.bias
F32
F32
[1280]
v.blk.9.attn_q.weight
F16
F16
[1280, 1280]
v.blk.9.attn_v.bias
F32
F32
[1280]
v.blk.9.attn_v.weight
F16
F16
[1280, 1280]
v.blk.9.ffn_down.bias
F32
F32
[1280]
v.blk.9.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.9.ffn_gate.bias
F32
F32
[3420]
v.blk.9.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.9.ffn_up.bias
F32
F32
[3420]
v.blk.9.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.9.ln1.weight
F32
F32
[1280]
v.blk.9.ln2.weight
F32
F32
[1280]
v.blk.10
v.blk.10.attn_k.bias
F32
F32
[1280]
v.blk.10.attn_k.weight
F16
F16
[1280, 1280]
v.blk.10.attn_out.bias
F32
F32
[1280]
v.blk.10.attn_out.weight
F16
F16
[1280, 1280]
v.blk.10.attn_q.bias
F32
F32
[1280]
v.blk.10.attn_q.weight
F16
F16
[1280, 1280]
v.blk.10.attn_v.bias
F32
F32
[1280]
v.blk.10.attn_v.weight
F16
F16
[1280, 1280]
v.blk.10.ffn_down.bias
F32
F32
[1280]
v.blk.10.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.10.ffn_gate.bias
F32
F32
[3420]
v.blk.10.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.10.ffn_up.bias
F32
F32
[3420]
v.blk.10.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.10.ln1.weight
F32
F32
[1280]
v.blk.10.ln2.weight
F32
F32
[1280]
v.blk.11
v.blk.11.attn_k.bias
F32
F32
[1280]
v.blk.11.attn_k.weight
F16
F16
[1280, 1280]
v.blk.11.attn_out.bias
F32
F32
[1280]
v.blk.11.attn_out.weight
F16
F16
[1280, 1280]
v.blk.11.attn_q.bias
F32
F32
[1280]
v.blk.11.attn_q.weight
F16
F16
[1280, 1280]
v.blk.11.attn_v.bias
F32
F32
[1280]
v.blk.11.attn_v.weight
F16
F16
[1280, 1280]
v.blk.11.ffn_down.bias
F32
F32
[1280]
v.blk.11.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.11.ffn_gate.bias
F32
F32
[3420]
v.blk.11.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.11.ffn_up.bias
F32
F32
[3420]
v.blk.11.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.11.ln1.weight
F32
F32
[1280]
v.blk.11.ln2.weight
F32
F32
[1280]
v.blk.12
v.blk.12.attn_k.bias
F32
F32
[1280]
v.blk.12.attn_k.weight
F16
F16
[1280, 1280]
v.blk.12.attn_out.bias
F32
F32
[1280]
v.blk.12.attn_out.weight
F16
F16
[1280, 1280]
v.blk.12.attn_q.bias
F32
F32
[1280]
v.blk.12.attn_q.weight
F16
F16
[1280, 1280]
v.blk.12.attn_v.bias
F32
F32
[1280]
v.blk.12.attn_v.weight
F16
F16
[1280, 1280]
v.blk.12.ffn_down.bias
F32
F32
[1280]
v.blk.12.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.12.ffn_gate.bias
F32
F32
[3420]
v.blk.12.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.12.ffn_up.bias
F32
F32
[3420]
v.blk.12.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.12.ln1.weight
F32
F32
[1280]
v.blk.12.ln2.weight
F32
F32
[1280]
v.blk.13
v.blk.13.attn_k.bias
F32
F32
[1280]
v.blk.13.attn_k.weight
F16
F16
[1280, 1280]
v.blk.13.attn_out.bias
F32
F32
[1280]
v.blk.13.attn_out.weight
F16
F16
[1280, 1280]
v.blk.13.attn_q.bias
F32
F32
[1280]
v.blk.13.attn_q.weight
F16
F16
[1280, 1280]
v.blk.13.attn_v.bias
F32
F32
[1280]
v.blk.13.attn_v.weight
F16
F16
[1280, 1280]
v.blk.13.ffn_down.bias
F32
F32
[1280]
v.blk.13.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.13.ffn_gate.bias
F32
F32
[3420]
v.blk.13.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.13.ffn_up.bias
F32
F32
[3420]
v.blk.13.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.13.ln1.weight
F32
F32
[1280]
v.blk.13.ln2.weight
F32
F32
[1280]
v.blk.14
v.blk.14.attn_k.bias
F32
F32
[1280]
v.blk.14.attn_k.weight
F16
F16
[1280, 1280]
v.blk.14.attn_out.bias
F32
F32
[1280]
v.blk.14.attn_out.weight
F16
F16
[1280, 1280]
v.blk.14.attn_q.bias
F32
F32
[1280]
v.blk.14.attn_q.weight
F16
F16
[1280, 1280]
v.blk.14.attn_v.bias
F32
F32
[1280]
v.blk.14.attn_v.weight
F16
F16
[1280, 1280]
v.blk.14.ffn_down.bias
F32
F32
[1280]
v.blk.14.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.14.ffn_gate.bias
F32
F32
[3420]
v.blk.14.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.14.ffn_up.bias
F32
F32
[3420]
v.blk.14.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.14.ln1.weight
F32
F32
[1280]
v.blk.14.ln2.weight
F32
F32
[1280]
v.blk.15
v.blk.15.attn_k.bias
F32
F32
[1280]
v.blk.15.attn_k.weight
F16
F16
[1280, 1280]
v.blk.15.attn_out.bias
F32
F32
[1280]
v.blk.15.attn_out.weight
F16
F16
[1280, 1280]
v.blk.15.attn_q.bias
F32
F32
[1280]
v.blk.15.attn_q.weight
F16
F16
[1280, 1280]
v.blk.15.attn_v.bias
F32
F32
[1280]
v.blk.15.attn_v.weight
F16
F16
[1280, 1280]
v.blk.15.ffn_down.bias
F32
F32
[1280]
v.blk.15.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.15.ffn_gate.bias
F32
F32
[3420]
v.blk.15.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.15.ffn_up.bias
F32
F32
[3420]
v.blk.15.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.15.ln1.weight
F32
F32
[1280]
v.blk.15.ln2.weight
F32
F32
[1280]
v.blk.16
v.blk.16.attn_k.bias
F32
F32
[1280]
v.blk.16.attn_k.weight
F16
F16
[1280, 1280]
v.blk.16.attn_out.bias
F32
F32
[1280]
v.blk.16.attn_out.weight
F16
F16
[1280, 1280]
v.blk.16.attn_q.bias
F32
F32
[1280]
v.blk.16.attn_q.weight
F16
F16
[1280, 1280]
v.blk.16.attn_v.bias
F32
F32
[1280]
v.blk.16.attn_v.weight
F16
F16
[1280, 1280]
v.blk.16.ffn_down.bias
F32
F32
[1280]
v.blk.16.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.16.ffn_gate.bias
F32
F32
[3420]
v.blk.16.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.16.ffn_up.bias
F32
F32
[3420]
v.blk.16.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.16.ln1.weight
F32
F32
[1280]
v.blk.16.ln2.weight
F32
F32
[1280]
v.blk.17
v.blk.17.attn_k.bias
F32
F32
[1280]
v.blk.17.attn_k.weight
F16
F16
[1280, 1280]
v.blk.17.attn_out.bias
F32
F32
[1280]
v.blk.17.attn_out.weight
F16
F16
[1280, 1280]
v.blk.17.attn_q.bias
F32
F32
[1280]
v.blk.17.attn_q.weight
F16
F16
[1280, 1280]
v.blk.17.attn_v.bias
F32
F32
[1280]
v.blk.17.attn_v.weight
F16
F16
[1280, 1280]
v.blk.17.ffn_down.bias
F32
F32
[1280]
v.blk.17.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.17.ffn_gate.bias
F32
F32
[3420]
v.blk.17.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.17.ffn_up.bias
F32
F32
[3420]
v.blk.17.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.17.ln1.weight
F32
F32
[1280]
v.blk.17.ln2.weight
F32
F32
[1280]
v.blk.18
v.blk.18.attn_k.bias
F32
F32
[1280]
v.blk.18.attn_k.weight
F16
F16
[1280, 1280]
v.blk.18.attn_out.bias
F32
F32
[1280]
v.blk.18.attn_out.weight
F16
F16
[1280, 1280]
v.blk.18.attn_q.bias
F32
F32
[1280]
v.blk.18.attn_q.weight
F16
F16
[1280, 1280]
v.blk.18.attn_v.bias
F32
F32
[1280]
v.blk.18.attn_v.weight
F16
F16
[1280, 1280]
v.blk.18.ffn_down.bias
F32
F32
[1280]
v.blk.18.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.18.ffn_gate.bias
F32
F32
[3420]
v.blk.18.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.18.ffn_up.bias
F32
F32
[3420]
v.blk.18.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.18.ln1.weight
F32
F32
[1280]
v.blk.18.ln2.weight
F32
F32
[1280]
v.blk.19
v.blk.19.attn_k.bias
F32
F32
[1280]
v.blk.19.attn_k.weight
F16
F16
[1280, 1280]
v.blk.19.attn_out.bias
F32
F32
[1280]
v.blk.19.attn_out.weight
F16
F16
[1280, 1280]
v.blk.19.attn_q.bias
F32
F32
[1280]
v.blk.19.attn_q.weight
F16
F16
[1280, 1280]
v.blk.19.attn_v.bias
F32
F32
[1280]
v.blk.19.attn_v.weight
F16
F16
[1280, 1280]
v.blk.19.ffn_down.bias
F32
F32
[1280]
v.blk.19.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.19.ffn_gate.bias
F32
F32
[3420]
v.blk.19.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.19.ffn_up.bias
F32
F32
[3420]
v.blk.19.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.19.ln1.weight
F32
F32
[1280]
v.blk.19.ln2.weight
F32
F32
[1280]
v.blk.20
v.blk.20.attn_k.bias
F32
F32
[1280]
v.blk.20.attn_k.weight
F16
F16
[1280, 1280]
v.blk.20.attn_out.bias
F32
F32
[1280]
v.blk.20.attn_out.weight
F16
F16
[1280, 1280]
v.blk.20.attn_q.bias
F32
F32
[1280]
v.blk.20.attn_q.weight
F16
F16
[1280, 1280]
v.blk.20.attn_v.bias
F32
F32
[1280]
v.blk.20.attn_v.weight
F16
F16
[1280, 1280]
v.blk.20.ffn_down.bias
F32
F32
[1280]
v.blk.20.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.20.ffn_gate.bias
F32
F32
[3420]
v.blk.20.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.20.ffn_up.bias
F32
F32
[3420]
v.blk.20.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.20.ln1.weight
F32
F32
[1280]
v.blk.20.ln2.weight
F32
F32
[1280]
v.blk.21
v.blk.21.attn_k.bias
F32
F32
[1280]
v.blk.21.attn_k.weight
F16
F16
[1280, 1280]
v.blk.21.attn_out.bias
F32
F32
[1280]
v.blk.21.attn_out.weight
F16
F16
[1280, 1280]
v.blk.21.attn_q.bias
F32
F32
[1280]
v.blk.21.attn_q.weight
F16
F16
[1280, 1280]
v.blk.21.attn_v.bias
F32
F32
[1280]
v.blk.21.attn_v.weight
F16
F16
[1280, 1280]
v.blk.21.ffn_down.bias
F32
F32
[1280]
v.blk.21.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.21.ffn_gate.bias
F32
F32
[3420]
v.blk.21.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.21.ffn_up.bias
F32
F32
[3420]
v.blk.21.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.21.ln1.weight
F32
F32
[1280]
v.blk.21.ln2.weight
F32
F32
[1280]
v.blk.22
v.blk.22.attn_k.bias
F32
F32
[1280]
v.blk.22.attn_k.weight
F16
F16
[1280, 1280]
v.blk.22.attn_out.bias
F32
F32
[1280]
v.blk.22.attn_out.weight
F16
F16
[1280, 1280]
v.blk.22.attn_q.bias
F32
F32
[1280]
v.blk.22.attn_q.weight
F16
F16
[1280, 1280]
v.blk.22.attn_v.bias
F32
F32
[1280]
v.blk.22.attn_v.weight
F16
F16
[1280, 1280]
v.blk.22.ffn_down.bias
F32
F32
[1280]
v.blk.22.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.22.ffn_gate.bias
F32
F32
[3420]
v.blk.22.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.22.ffn_up.bias
F32
F32
[3420]
v.blk.22.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.22.ln1.weight
F32
F32
[1280]
v.blk.22.ln2.weight
F32
F32
[1280]
v.blk.23
v.blk.23.attn_k.bias
F32
F32
[1280]
v.blk.23.attn_k.weight
F16
F16
[1280, 1280]
v.blk.23.attn_out.bias
F32
F32
[1280]
v.blk.23.attn_out.weight
F16
F16
[1280, 1280]
v.blk.23.attn_q.bias
F32
F32
[1280]
v.blk.23.attn_q.weight
F16
F16
[1280, 1280]
v.blk.23.attn_v.bias
F32
F32
[1280]
v.blk.23.attn_v.weight
F16
F16
[1280, 1280]
v.blk.23.ffn_down.bias
F32
F32
[1280]
v.blk.23.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.23.ffn_gate.bias
F32
F32
[3420]
v.blk.23.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.23.ffn_up.bias
F32
F32
[3420]
v.blk.23.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.23.ln1.weight
F32
F32
[1280]
v.blk.23.ln2.weight
F32
F32
[1280]
v.blk.24
v.blk.24.attn_k.bias
F32
F32
[1280]
v.blk.24.attn_k.weight
F16
F16
[1280, 1280]
v.blk.24.attn_out.bias
F32
F32
[1280]
v.blk.24.attn_out.weight
F16
F16
[1280, 1280]
v.blk.24.attn_q.bias
F32
F32
[1280]
v.blk.24.attn_q.weight
F16
F16
[1280, 1280]
v.blk.24.attn_v.bias
F32
F32
[1280]
v.blk.24.attn_v.weight
F16
F16
[1280, 1280]
v.blk.24.ffn_down.bias
F32
F32
[1280]
v.blk.24.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.24.ffn_gate.bias
F32
F32
[3420]
v.blk.24.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.24.ffn_up.bias
F32
F32
[3420]
v.blk.24.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.24.ln1.weight
F32
F32
[1280]
v.blk.24.ln2.weight
F32
F32
[1280]
v.blk.25
v.blk.25.attn_k.bias
F32
F32
[1280]
v.blk.25.attn_k.weight
F16
F16
[1280, 1280]
v.blk.25.attn_out.bias
F32
F32
[1280]
v.blk.25.attn_out.weight
F16
F16
[1280, 1280]
v.blk.25.attn_q.bias
F32
F32
[1280]
v.blk.25.attn_q.weight
F16
F16
[1280, 1280]
v.blk.25.attn_v.bias
F32
F32
[1280]
v.blk.25.attn_v.weight
F16
F16
[1280, 1280]
v.blk.25.ffn_down.bias
F32
F32
[1280]
v.blk.25.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.25.ffn_gate.bias
F32
F32
[3420]
v.blk.25.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.25.ffn_up.bias
F32
F32
[3420]
v.blk.25.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.25.ln1.weight
F32
F32
[1280]
v.blk.25.ln2.weight
F32
F32
[1280]
v.blk.26
v.blk.26.attn_k.bias
F32
F32
[1280]
v.blk.26.attn_k.weight
F16
F16
[1280, 1280]
v.blk.26.attn_out.bias
F32
F32
[1280]
v.blk.26.attn_out.weight
F16
F16
[1280, 1280]
v.blk.26.attn_q.bias
F32
F32
[1280]
v.blk.26.attn_q.weight
F16
F16
[1280, 1280]
v.blk.26.attn_v.bias
F32
F32
[1280]
v.blk.26.attn_v.weight
F16
F16
[1280, 1280]
v.blk.26.ffn_down.bias
F32
F32
[1280]
v.blk.26.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.26.ffn_gate.bias
F32
F32
[3420]
v.blk.26.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.26.ffn_up.bias
F32
F32
[3420]
v.blk.26.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.26.ln1.weight
F32
F32
[1280]
v.blk.26.ln2.weight
F32
F32
[1280]
v.blk.27
v.blk.27.attn_k.bias
F32
F32
[1280]
v.blk.27.attn_k.weight
F16
F16
[1280, 1280]
v.blk.27.attn_out.bias
F32
F32
[1280]
v.blk.27.attn_out.weight
F16
F16
[1280, 1280]
v.blk.27.attn_q.bias
F32
F32
[1280]
v.blk.27.attn_q.weight
F16
F16
[1280, 1280]
v.blk.27.attn_v.bias
F32
F32
[1280]
v.blk.27.attn_v.weight
F16
F16
[1280, 1280]
v.blk.27.ffn_down.bias
F32
F32
[1280]
v.blk.27.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.27.ffn_gate.bias
F32
F32
[3420]
v.blk.27.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.27.ffn_up.bias
F32
F32
[3420]
v.blk.27.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.27.ln1.weight
F32
F32
[1280]
v.blk.27.ln2.weight
F32
F32
[1280]
v.blk.28
v.blk.28.attn_k.bias
F32
F32
[1280]
v.blk.28.attn_k.weight
F16
F16
[1280, 1280]
v.blk.28.attn_out.bias
F32
F32
[1280]
v.blk.28.attn_out.weight
F16
F16
[1280, 1280]
v.blk.28.attn_q.bias
F32
F32
[1280]
v.blk.28.attn_q.weight
F16
F16
[1280, 1280]
v.blk.28.attn_v.bias
F32
F32
[1280]
v.blk.28.attn_v.weight
F16
F16
[1280, 1280]
v.blk.28.ffn_down.bias
F32
F32
[1280]
v.blk.28.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.28.ffn_gate.bias
F32
F32
[3420]
v.blk.28.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.28.ffn_up.bias
F32
F32
[3420]
v.blk.28.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.28.ln1.weight
F32
F32
[1280]
v.blk.28.ln2.weight
F32
F32
[1280]
v.blk.29
v.blk.29.attn_k.bias
F32
F32
[1280]
v.blk.29.attn_k.weight
F16
F16
[1280, 1280]
v.blk.29.attn_out.bias
F32
F32
[1280]
v.blk.29.attn_out.weight
F16
F16
[1280, 1280]
v.blk.29.attn_q.bias
F32
F32
[1280]
v.blk.29.attn_q.weight
F16
F16
[1280, 1280]
v.blk.29.attn_v.bias
F32
F32
[1280]
v.blk.29.attn_v.weight
F16
F16
[1280, 1280]
v.blk.29.ffn_down.bias
F32
F32
[1280]
v.blk.29.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.29.ffn_gate.bias
F32
F32
[3420]
v.blk.29.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.29.ffn_up.bias
F32
F32
[3420]
v.blk.29.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.29.ln1.weight
F32
F32
[1280]
v.blk.29.ln2.weight
F32
F32
[1280]
v.blk.30
v.blk.30.attn_k.bias
F32
F32
[1280]
v.blk.30.attn_k.weight
F16
F16
[1280, 1280]
v.blk.30.attn_out.bias
F32
F32
[1280]
v.blk.30.attn_out.weight
F16
F16
[1280, 1280]
v.blk.30.attn_q.bias
F32
F32
[1280]
v.blk.30.attn_q.weight
F16
F16
[1280, 1280]
v.blk.30.attn_v.bias
F32
F32
[1280]
v.blk.30.attn_v.weight
F16
F16
[1280, 1280]
v.blk.30.ffn_down.bias
F32
F32
[1280]
v.blk.30.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.30.ffn_gate.bias
F32
F32
[3420]
v.blk.30.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.30.ffn_up.bias
F32
F32
[3420]
v.blk.30.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.30.ln1.weight
F32
F32
[1280]
v.blk.30.ln2.weight
F32
F32
[1280]
v.blk.31
v.blk.31.attn_k.bias
F32
F32
[1280]
v.blk.31.attn_k.weight
F16
F16
[1280, 1280]
v.blk.31.attn_out.bias
F32
F32
[1280]
v.blk.31.attn_out.weight
F16
F16
[1280, 1280]
v.blk.31.attn_q.bias
F32
F32
[1280]
v.blk.31.attn_q.weight
F16
F16
[1280, 1280]
v.blk.31.attn_v.bias
F32
F32
[1280]
v.blk.31.attn_v.weight
F16
F16
[1280, 1280]
v.blk.31.ffn_down.bias
F32
F32
[1280]
v.blk.31.ffn_down.weight
F16
F16
[3420, 1280]
v.blk.31.ffn_gate.bias
F32
F32
[3420]
v.blk.31.ffn_gate.weight
F16
F16
[1280, 3420]
v.blk.31.ffn_up.bias
F32
F32
[3420]
v.blk.31.ffn_up.weight
F16
F16
[1280, 3420]
v.blk.31.ln1.weight
F32
F32
[1280]
v.blk.31.ln2.weight
F32
F32
[1280]
v.merger.ln_q.weight
F32
F32
[1280]
v.merger.mlp.0.bias
F32
F32
[5120]
v.merger.mlp.0.weight
F16
F16
[5120, 5120]
v.merger.mlp.2.bias
F32
F32
[2048]
v.merger.mlp.2.weight
F16
F16
[5120, 2048]
v.patch_embd_0.weight
F16
F16
[14, 14, 3, 1280]
v.patch_embd_1.weight
F16
F16
[14, 14, 3, 1280]
output_norm.weight
F32
F32
[2048]