GitHub
Discord
Docs
Cloud
Sign in
Download
Download
GitHub
Discord
Docs
Cloud
Sign in
Keyvan
/
german-ocr-turbo-fc
:latest
188
Downloads
Updated
1 month ago
Strukturierte Datenextraktion aus deutschen Dokumenten, Extrahiert Rechnungen, Formulare und Dokumente automatisch als JSON, bereit für Ihre Workflows.
Strukturierte Datenextraktion aus deutschen Dokumenten, Extrahiert Rechnungen, Formulare und Dokumente automatisch als JSON, bereit für Ihre Workflows.
Cancel
vision
tools
german-ocr-turbo-fc:latest
...
/
model
e4354d2ee7ea · 1.9GB
Metadata
general.architecture
qwen3vl
qwen3vl
general.file_type
Q4_K_M
Q4_K_M
qwen3vl.attention.head_count
16
16
qwen3vl.attention.head_count_kv
8
8
qwen3vl.attention.key_length
128
128
qwen3vl.attention.layer_norm_rms_epsilon
1e-06
1e-06
qwen3vl.attention.value_length
128
128
qwen3vl.block_count
28
28
qwen3vl.context_length
262144
262144
qwen3vl.embedding_length
2048
2048
qwen3vl.feed_forward_length
6144
6144
qwen3vl.rope.freq_base
5e+06
5e+06
qwen3vl.vision.attention.head_count
16
16
qwen3vl.vision.attention.layer_norm_epsilon
1e-06
1e-06
qwen3vl.vision.block_count
24
24
qwen3vl.vision.deepstack_visual_indexes
[5, 11, 17]
[5, 11, 17]
qwen3vl.vision.embedding_length
1024
1024
qwen3vl.vision.image_mean
[0.5, 0.5, 0.5]
[0.5, 0.5, 0.5]
qwen3vl.vision.image_std
[0.5, 0.5, 0.5]
[0.5, 0.5, 0.5]
qwen3vl.vision.longest_edge
16777216
16777216
qwen3vl.vision.num_channels
3
3
qwen3vl.vision.patch_size
16
16
qwen3vl.vision.rope.freq_base
10000
10000
qwen3vl.vision.shortest_edge
65536
65536
qwen3vl.vision.spatial_merge_size
2
2
qwen3vl.vision.temporal_patch_size
2
2
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
Q6_K
Q6_K
[2048, 151936]
blk.0
blk.0.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.0.attn_k_norm.weight
F32
F32
[128]
blk.0.attn_norm.weight
F32
F32
[2048]
blk.0.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.0.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.0.attn_q_norm.weight
F32
F32
[128]
blk.0.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.0.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.0.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.0.ffn_norm.weight
F32
F32
[2048]
blk.0.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.1
blk.1.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.1.attn_k_norm.weight
F32
F32
[128]
blk.1.attn_norm.weight
F32
F32
[2048]
blk.1.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.1.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.1.attn_q_norm.weight
F32
F32
[128]
blk.1.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.1.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.1.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.1.ffn_norm.weight
F32
F32
[2048]
blk.1.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.2
blk.2.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.2.attn_k_norm.weight
F32
F32
[128]
blk.2.attn_norm.weight
F32
F32
[2048]
blk.2.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.2.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.2.attn_q_norm.weight
F32
F32
[128]
blk.2.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.2.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.2.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.2.ffn_norm.weight
F32
F32
[2048]
blk.2.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.3
blk.3.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.3.attn_k_norm.weight
F32
F32
[128]
blk.3.attn_norm.weight
F32
F32
[2048]
blk.3.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.3.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.3.attn_q_norm.weight
F32
F32
[128]
blk.3.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.3.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.3.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.3.ffn_norm.weight
F32
F32
[2048]
blk.3.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.4
blk.4.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.4.attn_k_norm.weight
F32
F32
[128]
blk.4.attn_norm.weight
F32
F32
[2048]
blk.4.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.4.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.4.attn_q_norm.weight
F32
F32
[128]
blk.4.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.4.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.4.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.4.ffn_norm.weight
F32
F32
[2048]
blk.4.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.5
blk.5.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.5.attn_k_norm.weight
F32
F32
[128]
blk.5.attn_norm.weight
F32
F32
[2048]
blk.5.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.5.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.5.attn_q_norm.weight
F32
F32
[128]
blk.5.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.5.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.5.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.5.ffn_norm.weight
F32
F32
[2048]
blk.5.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.6
blk.6.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.6.attn_k_norm.weight
F32
F32
[128]
blk.6.attn_norm.weight
F32
F32
[2048]
blk.6.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.6.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.6.attn_q_norm.weight
F32
F32
[128]
blk.6.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.6.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.6.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.6.ffn_norm.weight
F32
F32
[2048]
blk.6.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.7
blk.7.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.7.attn_k_norm.weight
F32
F32
[128]
blk.7.attn_norm.weight
F32
F32
[2048]
blk.7.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.7.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.7.attn_q_norm.weight
F32
F32
[128]
blk.7.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.7.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.7.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.7.ffn_norm.weight
F32
F32
[2048]
blk.7.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.8
blk.8.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.8.attn_k_norm.weight
F32
F32
[128]
blk.8.attn_norm.weight
F32
F32
[2048]
blk.8.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.8.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.8.attn_q_norm.weight
F32
F32
[128]
blk.8.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.8.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.8.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.8.ffn_norm.weight
F32
F32
[2048]
blk.8.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.9
blk.9.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.9.attn_k_norm.weight
F32
F32
[128]
blk.9.attn_norm.weight
F32
F32
[2048]
blk.9.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.9.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.9.attn_q_norm.weight
F32
F32
[128]
blk.9.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.9.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.9.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.9.ffn_norm.weight
F32
F32
[2048]
blk.9.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.10
blk.10.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.10.attn_k_norm.weight
F32
F32
[128]
blk.10.attn_norm.weight
F32
F32
[2048]
blk.10.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.10.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.10.attn_q_norm.weight
F32
F32
[128]
blk.10.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.10.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.10.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.10.ffn_norm.weight
F32
F32
[2048]
blk.10.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.11
blk.11.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.11.attn_k_norm.weight
F32
F32
[128]
blk.11.attn_norm.weight
F32
F32
[2048]
blk.11.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.11.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.11.attn_q_norm.weight
F32
F32
[128]
blk.11.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.11.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.11.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.11.ffn_norm.weight
F32
F32
[2048]
blk.11.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.12
blk.12.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.12.attn_k_norm.weight
F32
F32
[128]
blk.12.attn_norm.weight
F32
F32
[2048]
blk.12.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.12.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.12.attn_q_norm.weight
F32
F32
[128]
blk.12.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.12.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.12.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.12.ffn_norm.weight
F32
F32
[2048]
blk.12.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.13
blk.13.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.13.attn_k_norm.weight
F32
F32
[128]
blk.13.attn_norm.weight
F32
F32
[2048]
blk.13.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.13.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.13.attn_q_norm.weight
F32
F32
[128]
blk.13.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.13.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.13.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.13.ffn_norm.weight
F32
F32
[2048]
blk.13.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.14
blk.14.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.14.attn_k_norm.weight
F32
F32
[128]
blk.14.attn_norm.weight
F32
F32
[2048]
blk.14.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.14.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.14.attn_q_norm.weight
F32
F32
[128]
blk.14.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.14.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.14.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.14.ffn_norm.weight
F32
F32
[2048]
blk.14.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.15
blk.15.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.15.attn_k_norm.weight
F32
F32
[128]
blk.15.attn_norm.weight
F32
F32
[2048]
blk.15.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.15.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.15.attn_q_norm.weight
F32
F32
[128]
blk.15.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.15.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.15.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.15.ffn_norm.weight
F32
F32
[2048]
blk.15.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.16
blk.16.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.16.attn_k_norm.weight
F32
F32
[128]
blk.16.attn_norm.weight
F32
F32
[2048]
blk.16.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.16.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.16.attn_q_norm.weight
F32
F32
[128]
blk.16.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.16.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.16.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.16.ffn_norm.weight
F32
F32
[2048]
blk.16.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.17
blk.17.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.17.attn_k_norm.weight
F32
F32
[128]
blk.17.attn_norm.weight
F32
F32
[2048]
blk.17.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.17.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.17.attn_q_norm.weight
F32
F32
[128]
blk.17.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.17.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.17.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.17.ffn_norm.weight
F32
F32
[2048]
blk.17.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.18
blk.18.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.18.attn_k_norm.weight
F32
F32
[128]
blk.18.attn_norm.weight
F32
F32
[2048]
blk.18.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.18.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.18.attn_q_norm.weight
F32
F32
[128]
blk.18.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.18.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.18.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.18.ffn_norm.weight
F32
F32
[2048]
blk.18.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.19
blk.19.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.19.attn_k_norm.weight
F32
F32
[128]
blk.19.attn_norm.weight
F32
F32
[2048]
blk.19.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.19.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.19.attn_q_norm.weight
F32
F32
[128]
blk.19.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.19.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.19.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.19.ffn_norm.weight
F32
F32
[2048]
blk.19.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.20
blk.20.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.20.attn_k_norm.weight
F32
F32
[128]
blk.20.attn_norm.weight
F32
F32
[2048]
blk.20.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.20.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.20.attn_q_norm.weight
F32
F32
[128]
blk.20.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.20.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.20.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.20.ffn_norm.weight
F32
F32
[2048]
blk.20.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.21
blk.21.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.21.attn_k_norm.weight
F32
F32
[128]
blk.21.attn_norm.weight
F32
F32
[2048]
blk.21.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.21.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.21.attn_q_norm.weight
F32
F32
[128]
blk.21.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.21.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.21.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.21.ffn_norm.weight
F32
F32
[2048]
blk.21.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.22
blk.22.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.22.attn_k_norm.weight
F32
F32
[128]
blk.22.attn_norm.weight
F32
F32
[2048]
blk.22.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.22.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.22.attn_q_norm.weight
F32
F32
[128]
blk.22.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.22.ffn_down.weight
Q4_K
Q4_K
[6144, 2048]
blk.22.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.22.ffn_norm.weight
F32
F32
[2048]
blk.22.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.23
blk.23.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.23.attn_k_norm.weight
F32
F32
[128]
blk.23.attn_norm.weight
F32
F32
[2048]
blk.23.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.23.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.23.attn_q_norm.weight
F32
F32
[128]
blk.23.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.23.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.23.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.23.ffn_norm.weight
F32
F32
[2048]
blk.23.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.24
blk.24.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.24.attn_k_norm.weight
F32
F32
[128]
blk.24.attn_norm.weight
F32
F32
[2048]
blk.24.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.24.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.24.attn_q_norm.weight
F32
F32
[128]
blk.24.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.24.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.24.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.24.ffn_norm.weight
F32
F32
[2048]
blk.24.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.25
blk.25.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.25.attn_k_norm.weight
F32
F32
[128]
blk.25.attn_norm.weight
F32
F32
[2048]
blk.25.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.25.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.25.attn_q_norm.weight
F32
F32
[128]
blk.25.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.25.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.25.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.25.ffn_norm.weight
F32
F32
[2048]
blk.25.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.26
blk.26.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.26.attn_k_norm.weight
F32
F32
[128]
blk.26.attn_norm.weight
F32
F32
[2048]
blk.26.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.26.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.26.attn_q_norm.weight
F32
F32
[128]
blk.26.attn_v.weight
Q6_K
Q6_K
[2048, 1024]
blk.26.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.26.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.26.ffn_norm.weight
F32
F32
[2048]
blk.26.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
blk.27
blk.27.attn_k.weight
Q4_K
Q4_K
[2048, 1024]
blk.27.attn_k_norm.weight
F32
F32
[128]
blk.27.attn_norm.weight
F32
F32
[2048]
blk.27.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.27.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.27.attn_q_norm.weight
F32
F32
[128]
blk.27.attn_v.weight
Q4_K
Q4_K
[2048, 1024]
blk.27.ffn_down.weight
Q6_K
Q6_K
[6144, 2048]
blk.27.ffn_gate.weight
Q4_K
Q4_K
[2048, 6144]
blk.27.ffn_norm.weight
F32
F32
[2048]
blk.27.ffn_up.weight
Q4_K
Q4_K
[2048, 6144]
v.blk.0
v.blk.0.attn_k.bias
F32
F32
[1024]
v.blk.0.attn_k.weight
F16
F16
[1024, 1024]
v.blk.0.attn_out.bias
F32
F32
[1024]
v.blk.0.attn_out.weight
F16
F16
[1024, 1024]
v.blk.0.attn_q.bias
F32
F32
[1024]
v.blk.0.attn_q.weight
F16
F16
[1024, 1024]
v.blk.0.attn_v.bias
F32
F32
[1024]
v.blk.0.attn_v.weight
Q4_K
Q4_K
[1024, 1024]
v.blk.0.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.0.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.0.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.0.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.0.norm1.bias
F32
F32
[1024]
v.blk.0.norm1.weight
F32
F32
[1024]
v.blk.0.norm2.bias
F32
F32
[1024]
v.blk.0.norm2.weight
F32
F32
[1024]
v.blk.1
v.blk.1.attn_k.bias
F32
F32
[1024]
v.blk.1.attn_k.weight
F16
F16
[1024, 1024]
v.blk.1.attn_out.bias
F32
F32
[1024]
v.blk.1.attn_out.weight
F16
F16
[1024, 1024]
v.blk.1.attn_q.bias
F32
F32
[1024]
v.blk.1.attn_q.weight
F16
F16
[1024, 1024]
v.blk.1.attn_v.bias
F32
F32
[1024]
v.blk.1.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.1.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.1.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.1.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.1.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.1.norm1.bias
F32
F32
[1024]
v.blk.1.norm1.weight
F32
F32
[1024]
v.blk.1.norm2.bias
F32
F32
[1024]
v.blk.1.norm2.weight
F32
F32
[1024]
v.blk.2
v.blk.2.attn_k.bias
F32
F32
[1024]
v.blk.2.attn_k.weight
F16
F16
[1024, 1024]
v.blk.2.attn_out.bias
F32
F32
[1024]
v.blk.2.attn_out.weight
F16
F16
[1024, 1024]
v.blk.2.attn_q.bias
F32
F32
[1024]
v.blk.2.attn_q.weight
F16
F16
[1024, 1024]
v.blk.2.attn_v.bias
F32
F32
[1024]
v.blk.2.attn_v.weight
Q4_K
Q4_K
[1024, 1024]
v.blk.2.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.2.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.2.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.2.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.2.norm1.bias
F32
F32
[1024]
v.blk.2.norm1.weight
F32
F32
[1024]
v.blk.2.norm2.bias
F32
F32
[1024]
v.blk.2.norm2.weight
F32
F32
[1024]
v.blk.3
v.blk.3.attn_k.bias
F32
F32
[1024]
v.blk.3.attn_k.weight
F16
F16
[1024, 1024]
v.blk.3.attn_out.bias
F32
F32
[1024]
v.blk.3.attn_out.weight
F16
F16
[1024, 1024]
v.blk.3.attn_q.bias
F32
F32
[1024]
v.blk.3.attn_q.weight
F16
F16
[1024, 1024]
v.blk.3.attn_v.bias
F32
F32
[1024]
v.blk.3.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.3.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.3.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.3.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.3.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.3.norm1.bias
F32
F32
[1024]
v.blk.3.norm1.weight
F32
F32
[1024]
v.blk.3.norm2.bias
F32
F32
[1024]
v.blk.3.norm2.weight
F32
F32
[1024]
v.blk.4
v.blk.4.attn_k.bias
F32
F32
[1024]
v.blk.4.attn_k.weight
F16
F16
[1024, 1024]
v.blk.4.attn_out.bias
F32
F32
[1024]
v.blk.4.attn_out.weight
F16
F16
[1024, 1024]
v.blk.4.attn_q.bias
F32
F32
[1024]
v.blk.4.attn_q.weight
F16
F16
[1024, 1024]
v.blk.4.attn_v.bias
F32
F32
[1024]
v.blk.4.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.4.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.4.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.4.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.4.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.4.norm1.bias
F32
F32
[1024]
v.blk.4.norm1.weight
F32
F32
[1024]
v.blk.4.norm2.bias
F32
F32
[1024]
v.blk.4.norm2.weight
F32
F32
[1024]
v.blk.5
v.blk.5.attn_k.bias
F32
F32
[1024]
v.blk.5.attn_k.weight
F16
F16
[1024, 1024]
v.blk.5.attn_out.bias
F32
F32
[1024]
v.blk.5.attn_out.weight
F16
F16
[1024, 1024]
v.blk.5.attn_q.bias
F32
F32
[1024]
v.blk.5.attn_q.weight
F16
F16
[1024, 1024]
v.blk.5.attn_v.bias
F32
F32
[1024]
v.blk.5.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.5.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.5.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.5.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.5.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.5.norm1.bias
F32
F32
[1024]
v.blk.5.norm1.weight
F32
F32
[1024]
v.blk.5.norm2.bias
F32
F32
[1024]
v.blk.5.norm2.weight
F32
F32
[1024]
v.blk.6
v.blk.6.attn_k.bias
F32
F32
[1024]
v.blk.6.attn_k.weight
F16
F16
[1024, 1024]
v.blk.6.attn_out.bias
F32
F32
[1024]
v.blk.6.attn_out.weight
F16
F16
[1024, 1024]
v.blk.6.attn_q.bias
F32
F32
[1024]
v.blk.6.attn_q.weight
F16
F16
[1024, 1024]
v.blk.6.attn_v.bias
F32
F32
[1024]
v.blk.6.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.6.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.6.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.6.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.6.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.6.norm1.bias
F32
F32
[1024]
v.blk.6.norm1.weight
F32
F32
[1024]
v.blk.6.norm2.bias
F32
F32
[1024]
v.blk.6.norm2.weight
F32
F32
[1024]
v.blk.7
v.blk.7.attn_k.bias
F32
F32
[1024]
v.blk.7.attn_k.weight
F16
F16
[1024, 1024]
v.blk.7.attn_out.bias
F32
F32
[1024]
v.blk.7.attn_out.weight
F16
F16
[1024, 1024]
v.blk.7.attn_q.bias
F32
F32
[1024]
v.blk.7.attn_q.weight
F16
F16
[1024, 1024]
v.blk.7.attn_v.bias
F32
F32
[1024]
v.blk.7.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.7.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.7.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.7.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.7.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.7.norm1.bias
F32
F32
[1024]
v.blk.7.norm1.weight
F32
F32
[1024]
v.blk.7.norm2.bias
F32
F32
[1024]
v.blk.7.norm2.weight
F32
F32
[1024]
v.blk.8
v.blk.8.attn_k.bias
F32
F32
[1024]
v.blk.8.attn_k.weight
F16
F16
[1024, 1024]
v.blk.8.attn_out.bias
F32
F32
[1024]
v.blk.8.attn_out.weight
F16
F16
[1024, 1024]
v.blk.8.attn_q.bias
F32
F32
[1024]
v.blk.8.attn_q.weight
F16
F16
[1024, 1024]
v.blk.8.attn_v.bias
F32
F32
[1024]
v.blk.8.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.8.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.8.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.8.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.8.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.8.norm1.bias
F32
F32
[1024]
v.blk.8.norm1.weight
F32
F32
[1024]
v.blk.8.norm2.bias
F32
F32
[1024]
v.blk.8.norm2.weight
F32
F32
[1024]
v.blk.9
v.blk.9.attn_k.bias
F32
F32
[1024]
v.blk.9.attn_k.weight
F16
F16
[1024, 1024]
v.blk.9.attn_out.bias
F32
F32
[1024]
v.blk.9.attn_out.weight
F16
F16
[1024, 1024]
v.blk.9.attn_q.bias
F32
F32
[1024]
v.blk.9.attn_q.weight
F16
F16
[1024, 1024]
v.blk.9.attn_v.bias
F32
F32
[1024]
v.blk.9.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.9.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.9.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.9.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.9.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.9.norm1.bias
F32
F32
[1024]
v.blk.9.norm1.weight
F32
F32
[1024]
v.blk.9.norm2.bias
F32
F32
[1024]
v.blk.9.norm2.weight
F32
F32
[1024]
v.blk.10
v.blk.10.attn_k.bias
F32
F32
[1024]
v.blk.10.attn_k.weight
F16
F16
[1024, 1024]
v.blk.10.attn_out.bias
F32
F32
[1024]
v.blk.10.attn_out.weight
F16
F16
[1024, 1024]
v.blk.10.attn_q.bias
F32
F32
[1024]
v.blk.10.attn_q.weight
F16
F16
[1024, 1024]
v.blk.10.attn_v.bias
F32
F32
[1024]
v.blk.10.attn_v.weight
Q4_K
Q4_K
[1024, 1024]
v.blk.10.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.10.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.10.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.10.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.10.norm1.bias
F32
F32
[1024]
v.blk.10.norm1.weight
F32
F32
[1024]
v.blk.10.norm2.bias
F32
F32
[1024]
v.blk.10.norm2.weight
F32
F32
[1024]
v.blk.11
v.blk.11.attn_k.bias
F32
F32
[1024]
v.blk.11.attn_k.weight
F16
F16
[1024, 1024]
v.blk.11.attn_out.bias
F32
F32
[1024]
v.blk.11.attn_out.weight
F16
F16
[1024, 1024]
v.blk.11.attn_q.bias
F32
F32
[1024]
v.blk.11.attn_q.weight
F16
F16
[1024, 1024]
v.blk.11.attn_v.bias
F32
F32
[1024]
v.blk.11.attn_v.weight
Q4_K
Q4_K
[1024, 1024]
v.blk.11.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.11.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.11.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.11.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.11.norm1.bias
F32
F32
[1024]
v.blk.11.norm1.weight
F32
F32
[1024]
v.blk.11.norm2.bias
F32
F32
[1024]
v.blk.11.norm2.weight
F32
F32
[1024]
v.blk.12
v.blk.12.attn_k.bias
F32
F32
[1024]
v.blk.12.attn_k.weight
F16
F16
[1024, 1024]
v.blk.12.attn_out.bias
F32
F32
[1024]
v.blk.12.attn_out.weight
F16
F16
[1024, 1024]
v.blk.12.attn_q.bias
F32
F32
[1024]
v.blk.12.attn_q.weight
F16
F16
[1024, 1024]
v.blk.12.attn_v.bias
F32
F32
[1024]
v.blk.12.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.12.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.12.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.12.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.12.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.12.norm1.bias
F32
F32
[1024]
v.blk.12.norm1.weight
F32
F32
[1024]
v.blk.12.norm2.bias
F32
F32
[1024]
v.blk.12.norm2.weight
F32
F32
[1024]
v.blk.13
v.blk.13.attn_k.bias
F32
F32
[1024]
v.blk.13.attn_k.weight
F16
F16
[1024, 1024]
v.blk.13.attn_out.bias
F32
F32
[1024]
v.blk.13.attn_out.weight
F16
F16
[1024, 1024]
v.blk.13.attn_q.bias
F32
F32
[1024]
v.blk.13.attn_q.weight
F16
F16
[1024, 1024]
v.blk.13.attn_v.bias
F32
F32
[1024]
v.blk.13.attn_v.weight
Q4_K
Q4_K
[1024, 1024]
v.blk.13.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.13.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.13.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.13.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.13.norm1.bias
F32
F32
[1024]
v.blk.13.norm1.weight
F32
F32
[1024]
v.blk.13.norm2.bias
F32
F32
[1024]
v.blk.13.norm2.weight
F32
F32
[1024]
v.blk.14
v.blk.14.attn_k.bias
F32
F32
[1024]
v.blk.14.attn_k.weight
F16
F16
[1024, 1024]
v.blk.14.attn_out.bias
F32
F32
[1024]
v.blk.14.attn_out.weight
F16
F16
[1024, 1024]
v.blk.14.attn_q.bias
F32
F32
[1024]
v.blk.14.attn_q.weight
F16
F16
[1024, 1024]
v.blk.14.attn_v.bias
F32
F32
[1024]
v.blk.14.attn_v.weight
Q4_K
Q4_K
[1024, 1024]
v.blk.14.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.14.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.14.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.14.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.14.norm1.bias
F32
F32
[1024]
v.blk.14.norm1.weight
F32
F32
[1024]
v.blk.14.norm2.bias
F32
F32
[1024]
v.blk.14.norm2.weight
F32
F32
[1024]
v.blk.15
v.blk.15.attn_k.bias
F32
F32
[1024]
v.blk.15.attn_k.weight
F16
F16
[1024, 1024]
v.blk.15.attn_out.bias
F32
F32
[1024]
v.blk.15.attn_out.weight
F16
F16
[1024, 1024]
v.blk.15.attn_q.bias
F32
F32
[1024]
v.blk.15.attn_q.weight
F16
F16
[1024, 1024]
v.blk.15.attn_v.bias
F32
F32
[1024]
v.blk.15.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.15.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.15.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.15.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.15.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.15.norm1.bias
F32
F32
[1024]
v.blk.15.norm1.weight
F32
F32
[1024]
v.blk.15.norm2.bias
F32
F32
[1024]
v.blk.15.norm2.weight
F32
F32
[1024]
v.blk.16
v.blk.16.attn_k.bias
F32
F32
[1024]
v.blk.16.attn_k.weight
F16
F16
[1024, 1024]
v.blk.16.attn_out.bias
F32
F32
[1024]
v.blk.16.attn_out.weight
F16
F16
[1024, 1024]
v.blk.16.attn_q.bias
F32
F32
[1024]
v.blk.16.attn_q.weight
F16
F16
[1024, 1024]
v.blk.16.attn_v.bias
F32
F32
[1024]
v.blk.16.attn_v.weight
Q4_K
Q4_K
[1024, 1024]
v.blk.16.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.16.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.16.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.16.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.16.norm1.bias
F32
F32
[1024]
v.blk.16.norm1.weight
F32
F32
[1024]
v.blk.16.norm2.bias
F32
F32
[1024]
v.blk.16.norm2.weight
F32
F32
[1024]
v.blk.17
v.blk.17.attn_k.bias
F32
F32
[1024]
v.blk.17.attn_k.weight
F16
F16
[1024, 1024]
v.blk.17.attn_out.bias
F32
F32
[1024]
v.blk.17.attn_out.weight
F16
F16
[1024, 1024]
v.blk.17.attn_q.bias
F32
F32
[1024]
v.blk.17.attn_q.weight
F16
F16
[1024, 1024]
v.blk.17.attn_v.bias
F32
F32
[1024]
v.blk.17.attn_v.weight
Q4_K
Q4_K
[1024, 1024]
v.blk.17.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.17.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.17.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.17.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.17.norm1.bias
F32
F32
[1024]
v.blk.17.norm1.weight
F32
F32
[1024]
v.blk.17.norm2.bias
F32
F32
[1024]
v.blk.17.norm2.weight
F32
F32
[1024]
v.blk.18
v.blk.18.attn_k.bias
F32
F32
[1024]
v.blk.18.attn_k.weight
F16
F16
[1024, 1024]
v.blk.18.attn_out.bias
F32
F32
[1024]
v.blk.18.attn_out.weight
F16
F16
[1024, 1024]
v.blk.18.attn_q.bias
F32
F32
[1024]
v.blk.18.attn_q.weight
F16
F16
[1024, 1024]
v.blk.18.attn_v.bias
F32
F32
[1024]
v.blk.18.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.18.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.18.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.18.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.18.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.18.norm1.bias
F32
F32
[1024]
v.blk.18.norm1.weight
F32
F32
[1024]
v.blk.18.norm2.bias
F32
F32
[1024]
v.blk.18.norm2.weight
F32
F32
[1024]
v.blk.19
v.blk.19.attn_k.bias
F32
F32
[1024]
v.blk.19.attn_k.weight
F16
F16
[1024, 1024]
v.blk.19.attn_out.bias
F32
F32
[1024]
v.blk.19.attn_out.weight
F16
F16
[1024, 1024]
v.blk.19.attn_q.bias
F32
F32
[1024]
v.blk.19.attn_q.weight
F16
F16
[1024, 1024]
v.blk.19.attn_v.bias
F32
F32
[1024]
v.blk.19.attn_v.weight
Q4_K
Q4_K
[1024, 1024]
v.blk.19.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.19.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.19.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.19.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.19.norm1.bias
F32
F32
[1024]
v.blk.19.norm1.weight
F32
F32
[1024]
v.blk.19.norm2.bias
F32
F32
[1024]
v.blk.19.norm2.weight
F32
F32
[1024]
v.blk.20
v.blk.20.attn_k.bias
F32
F32
[1024]
v.blk.20.attn_k.weight
F16
F16
[1024, 1024]
v.blk.20.attn_out.bias
F32
F32
[1024]
v.blk.20.attn_out.weight
F16
F16
[1024, 1024]
v.blk.20.attn_q.bias
F32
F32
[1024]
v.blk.20.attn_q.weight
F16
F16
[1024, 1024]
v.blk.20.attn_v.bias
F32
F32
[1024]
v.blk.20.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.20.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.20.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.20.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.20.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.20.norm1.bias
F32
F32
[1024]
v.blk.20.norm1.weight
F32
F32
[1024]
v.blk.20.norm2.bias
F32
F32
[1024]
v.blk.20.norm2.weight
F32
F32
[1024]
v.blk.21
v.blk.21.attn_k.bias
F32
F32
[1024]
v.blk.21.attn_k.weight
F16
F16
[1024, 1024]
v.blk.21.attn_out.bias
F32
F32
[1024]
v.blk.21.attn_out.weight
F16
F16
[1024, 1024]
v.blk.21.attn_q.bias
F32
F32
[1024]
v.blk.21.attn_q.weight
F16
F16
[1024, 1024]
v.blk.21.attn_v.bias
F32
F32
[1024]
v.blk.21.attn_v.weight
Q4_K
Q4_K
[1024, 1024]
v.blk.21.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.21.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.21.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.21.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.21.norm1.bias
F32
F32
[1024]
v.blk.21.norm1.weight
F32
F32
[1024]
v.blk.21.norm2.bias
F32
F32
[1024]
v.blk.21.norm2.weight
F32
F32
[1024]
v.blk.22
v.blk.22.attn_k.bias
F32
F32
[1024]
v.blk.22.attn_k.weight
F16
F16
[1024, 1024]
v.blk.22.attn_out.bias
F32
F32
[1024]
v.blk.22.attn_out.weight
F16
F16
[1024, 1024]
v.blk.22.attn_q.bias
F32
F32
[1024]
v.blk.22.attn_q.weight
F16
F16
[1024, 1024]
v.blk.22.attn_v.bias
F32
F32
[1024]
v.blk.22.attn_v.weight
Q4_K
Q4_K
[1024, 1024]
v.blk.22.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.22.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.22.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.22.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.22.norm1.bias
F32
F32
[1024]
v.blk.22.norm1.weight
F32
F32
[1024]
v.blk.22.norm2.bias
F32
F32
[1024]
v.blk.22.norm2.weight
F32
F32
[1024]
v.blk.23
v.blk.23.attn_k.bias
F32
F32
[1024]
v.blk.23.attn_k.weight
F16
F16
[1024, 1024]
v.blk.23.attn_out.bias
F32
F32
[1024]
v.blk.23.attn_out.weight
F16
F16
[1024, 1024]
v.blk.23.attn_q.bias
F32
F32
[1024]
v.blk.23.attn_q.weight
F16
F16
[1024, 1024]
v.blk.23.attn_v.bias
F32
F32
[1024]
v.blk.23.attn_v.weight
Q6_K
Q6_K
[1024, 1024]
v.blk.23.mlp.linear_fc1.bias
F32
F32
[4096]
v.blk.23.mlp.linear_fc1.weight
F16
F16
[1024, 4096]
v.blk.23.mlp.linear_fc2.bias
F32
F32
[1024]
v.blk.23.mlp.linear_fc2.weight
F16
F16
[4096, 1024]
v.blk.23.norm1.bias
F32
F32
[1024]
v.blk.23.norm1.weight
F32
F32
[1024]
v.blk.23.norm2.bias
F32
F32
[1024]
v.blk.23.norm2.weight
F32
F32
[1024]
v.deepstack_merger.0.linear_fc1.bias
F32
F32
[4096]
v.deepstack_merger.0.linear_fc1.weight
F16
F16
[4096, 4096]
v.deepstack_merger.0.linear_fc2.bias
F32
F32
[2048]
v.deepstack_merger.0.linear_fc2.weight
F16
F16
[4096, 2048]
v.deepstack_merger.0.norm.bias
F32
F32
[4096]
v.deepstack_merger.0.norm.weight
F32
F32
[4096]
v.deepstack_merger.1.linear_fc1.bias
F32
F32
[4096]
v.deepstack_merger.1.linear_fc1.weight
F16
F16
[4096, 4096]
v.deepstack_merger.1.linear_fc2.bias
F32
F32
[2048]
v.deepstack_merger.1.linear_fc2.weight
F16
F16
[4096, 2048]
v.deepstack_merger.1.norm.bias
F32
F32
[4096]
v.deepstack_merger.1.norm.weight
F32
F32
[4096]
v.deepstack_merger.2.linear_fc1.bias
F32
F32
[4096]
v.deepstack_merger.2.linear_fc1.weight
F16
F16
[4096, 4096]
v.deepstack_merger.2.linear_fc2.bias
F32
F32
[2048]
v.deepstack_merger.2.linear_fc2.weight
F16
F16
[4096, 2048]
v.deepstack_merger.2.norm.bias
F32
F32
[4096]
v.deepstack_merger.2.norm.weight
F32
F32
[4096]
v.merger.linear_fc1.bias
F32
F32
[4096]
v.merger.linear_fc1.weight
F16
F16
[4096, 4096]
v.merger.linear_fc2.bias
F32
F32
[2048]
v.merger.linear_fc2.weight
F16
F16
[4096, 2048]
v.merger.norm.bias
F32
F32
[1024]
v.merger.norm.weight
F32
F32
[1024]
v.patch_embed.bias
F32
F32
[1024]
v.patch_embed.weight
F16
F16
[16, 16, 2, 3072]
v.pos_embed.weight
F16
F16
[1024, 2304]
output_norm.weight
F32
F32
[2048]