Models
GitHub
Discord
Docs
Cloud
Sign in
Download
Models
Download
GitHub
Discord
Docs
Cloud
Sign in
mranv
/
ollash
:latest
59
Downloads
Updated
1 month ago
An intelligent shell command assistant specialized in generating accurate, safe, and efficient shell commands for various operating systems (Linux)
An intelligent shell command assistant specialized in generating accurate, safe, and efficient shell commands for various operating systems (Linux)
Cancel
ollash:latest
...
/
model
0e9f4fe7127a · 1.6GB
Metadata
general.architecture
qwen2
qwen2
general.file_type
Q8_0
Q8_0
qwen2.attention.head_count
12
12
qwen2.attention.head_count_kv
2
2
qwen2.attention.layer_norm_rms_epsilon
1e-06
1e-06
qwen2.block_count
28
28
qwen2.context_length
131072
131072
qwen2.embedding_length
1536
1536
qwen2.feed_forward_length
8960
8960
qwen2.rope.freq_base
1e+06
1e+06
tokenizer.ggml.bos_token_id
151643
151643
tokenizer.ggml.eos_token_id
151643
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.token_type
[1, 1, 1, 1, 1, ...]
[1, 1, 1, 1, 1, ...]
tokenizer.ggml.tokens
[!, ", #, $, %, ...]
[!, ", #, $, %, ...]
Tensor
Name
Type
Shape
token_embd.weight
Q8_0
Q8_0
[1536, 151936]
blk.0
blk.0.attn_k.bias
F32
F32
[256]
blk.0.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.0.attn_norm.weight
F32
F32
[1536]
blk.0.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.0.attn_q.bias
F32
F32
[1536]
blk.0.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.0.attn_v.bias
F32
F32
[256]
blk.0.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.0.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.0.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.0.ffn_norm.weight
F32
F32
[1536]
blk.0.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.1
blk.1.attn_k.bias
F32
F32
[256]
blk.1.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.1.attn_norm.weight
F32
F32
[1536]
blk.1.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.1.attn_q.bias
F32
F32
[1536]
blk.1.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.1.attn_v.bias
F32
F32
[256]
blk.1.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.1.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.1.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.1.ffn_norm.weight
F32
F32
[1536]
blk.1.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.2
blk.2.attn_k.bias
F32
F32
[256]
blk.2.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.2.attn_norm.weight
F32
F32
[1536]
blk.2.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.2.attn_q.bias
F32
F32
[1536]
blk.2.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.2.attn_v.bias
F32
F32
[256]
blk.2.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.2.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.2.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.2.ffn_norm.weight
F32
F32
[1536]
blk.2.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.3
blk.3.attn_k.bias
F32
F32
[256]
blk.3.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.3.attn_norm.weight
F32
F32
[1536]
blk.3.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.3.attn_q.bias
F32
F32
[1536]
blk.3.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.3.attn_v.bias
F32
F32
[256]
blk.3.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.3.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.3.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.3.ffn_norm.weight
F32
F32
[1536]
blk.3.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.4
blk.4.attn_k.bias
F32
F32
[256]
blk.4.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.4.attn_norm.weight
F32
F32
[1536]
blk.4.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.4.attn_q.bias
F32
F32
[1536]
blk.4.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.4.attn_v.bias
F32
F32
[256]
blk.4.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.4.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.4.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.4.ffn_norm.weight
F32
F32
[1536]
blk.4.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.5
blk.5.attn_k.bias
F32
F32
[256]
blk.5.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.5.attn_norm.weight
F32
F32
[1536]
blk.5.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.5.attn_q.bias
F32
F32
[1536]
blk.5.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.5.attn_v.bias
F32
F32
[256]
blk.5.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.5.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.5.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.5.ffn_norm.weight
F32
F32
[1536]
blk.5.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.6
blk.6.attn_k.bias
F32
F32
[256]
blk.6.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.6.attn_norm.weight
F32
F32
[1536]
blk.6.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.6.attn_q.bias
F32
F32
[1536]
blk.6.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.6.attn_v.bias
F32
F32
[256]
blk.6.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.6.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.6.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.6.ffn_norm.weight
F32
F32
[1536]
blk.6.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.7
blk.7.attn_k.bias
F32
F32
[256]
blk.7.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.7.attn_norm.weight
F32
F32
[1536]
blk.7.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.7.attn_q.bias
F32
F32
[1536]
blk.7.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.7.attn_v.bias
F32
F32
[256]
blk.7.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.7.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.7.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.7.ffn_norm.weight
F32
F32
[1536]
blk.7.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.8
blk.8.attn_k.bias
F32
F32
[256]
blk.8.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.8.attn_norm.weight
F32
F32
[1536]
blk.8.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.8.attn_q.bias
F32
F32
[1536]
blk.8.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.8.attn_v.bias
F32
F32
[256]
blk.8.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.8.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.8.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.8.ffn_norm.weight
F32
F32
[1536]
blk.8.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.9
blk.9.attn_k.bias
F32
F32
[256]
blk.9.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.9.attn_norm.weight
F32
F32
[1536]
blk.9.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.9.attn_q.bias
F32
F32
[1536]
blk.9.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.9.attn_v.bias
F32
F32
[256]
blk.9.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.9.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.9.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.9.ffn_norm.weight
F32
F32
[1536]
blk.9.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.10
blk.10.attn_k.bias
F32
F32
[256]
blk.10.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.10.attn_norm.weight
F32
F32
[1536]
blk.10.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.10.attn_q.bias
F32
F32
[1536]
blk.10.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.10.attn_v.bias
F32
F32
[256]
blk.10.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.10.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.10.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.10.ffn_norm.weight
F32
F32
[1536]
blk.10.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.11
blk.11.attn_k.bias
F32
F32
[256]
blk.11.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.11.attn_norm.weight
F32
F32
[1536]
blk.11.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.11.attn_q.bias
F32
F32
[1536]
blk.11.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.11.attn_v.bias
F32
F32
[256]
blk.11.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.11.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.11.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.11.ffn_norm.weight
F32
F32
[1536]
blk.11.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.12
blk.12.attn_k.bias
F32
F32
[256]
blk.12.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.12.attn_norm.weight
F32
F32
[1536]
blk.12.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.12.attn_q.bias
F32
F32
[1536]
blk.12.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.12.attn_v.bias
F32
F32
[256]
blk.12.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.12.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.12.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.12.ffn_norm.weight
F32
F32
[1536]
blk.12.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.13
blk.13.attn_k.bias
F32
F32
[256]
blk.13.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.13.attn_norm.weight
F32
F32
[1536]
blk.13.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.13.attn_q.bias
F32
F32
[1536]
blk.13.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.13.attn_v.bias
F32
F32
[256]
blk.13.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.13.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.13.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.13.ffn_norm.weight
F32
F32
[1536]
blk.13.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.14
blk.14.attn_k.bias
F32
F32
[256]
blk.14.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.14.attn_norm.weight
F32
F32
[1536]
blk.14.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.14.attn_q.bias
F32
F32
[1536]
blk.14.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.14.attn_v.bias
F32
F32
[256]
blk.14.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.14.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.14.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.14.ffn_norm.weight
F32
F32
[1536]
blk.14.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.15
blk.15.attn_k.bias
F32
F32
[256]
blk.15.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.15.attn_norm.weight
F32
F32
[1536]
blk.15.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.15.attn_q.bias
F32
F32
[1536]
blk.15.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.15.attn_v.bias
F32
F32
[256]
blk.15.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.15.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.15.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.15.ffn_norm.weight
F32
F32
[1536]
blk.15.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.16
blk.16.attn_k.bias
F32
F32
[256]
blk.16.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.16.attn_norm.weight
F32
F32
[1536]
blk.16.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.16.attn_q.bias
F32
F32
[1536]
blk.16.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.16.attn_v.bias
F32
F32
[256]
blk.16.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.16.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.16.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.16.ffn_norm.weight
F32
F32
[1536]
blk.16.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.17
blk.17.attn_k.bias
F32
F32
[256]
blk.17.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.17.attn_norm.weight
F32
F32
[1536]
blk.17.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.17.attn_q.bias
F32
F32
[1536]
blk.17.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.17.attn_v.bias
F32
F32
[256]
blk.17.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.17.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.17.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.17.ffn_norm.weight
F32
F32
[1536]
blk.17.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.18
blk.18.attn_k.bias
F32
F32
[256]
blk.18.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.18.attn_norm.weight
F32
F32
[1536]
blk.18.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.18.attn_q.bias
F32
F32
[1536]
blk.18.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.18.attn_v.bias
F32
F32
[256]
blk.18.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.18.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.18.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.18.ffn_norm.weight
F32
F32
[1536]
blk.18.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.19
blk.19.attn_k.bias
F32
F32
[256]
blk.19.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.19.attn_norm.weight
F32
F32
[1536]
blk.19.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.19.attn_q.bias
F32
F32
[1536]
blk.19.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.19.attn_v.bias
F32
F32
[256]
blk.19.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.19.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.19.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.19.ffn_norm.weight
F32
F32
[1536]
blk.19.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.20
blk.20.attn_k.bias
F32
F32
[256]
blk.20.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.20.attn_norm.weight
F32
F32
[1536]
blk.20.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.20.attn_q.bias
F32
F32
[1536]
blk.20.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.20.attn_v.bias
F32
F32
[256]
blk.20.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.20.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.20.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.20.ffn_norm.weight
F32
F32
[1536]
blk.20.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.21
blk.21.attn_k.bias
F32
F32
[256]
blk.21.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.21.attn_norm.weight
F32
F32
[1536]
blk.21.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.21.attn_q.bias
F32
F32
[1536]
blk.21.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.21.attn_v.bias
F32
F32
[256]
blk.21.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.21.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.21.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.21.ffn_norm.weight
F32
F32
[1536]
blk.21.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.22
blk.22.attn_k.bias
F32
F32
[256]
blk.22.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.22.attn_norm.weight
F32
F32
[1536]
blk.22.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.22.attn_q.bias
F32
F32
[1536]
blk.22.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.22.attn_v.bias
F32
F32
[256]
blk.22.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.22.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.22.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.22.ffn_norm.weight
F32
F32
[1536]
blk.22.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.23
blk.23.attn_k.bias
F32
F32
[256]
blk.23.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.23.attn_norm.weight
F32
F32
[1536]
blk.23.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.23.attn_q.bias
F32
F32
[1536]
blk.23.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.23.attn_v.bias
F32
F32
[256]
blk.23.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.23.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.23.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.23.ffn_norm.weight
F32
F32
[1536]
blk.23.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.24
blk.24.attn_k.bias
F32
F32
[256]
blk.24.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.24.attn_norm.weight
F32
F32
[1536]
blk.24.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.24.attn_q.bias
F32
F32
[1536]
blk.24.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.24.attn_v.bias
F32
F32
[256]
blk.24.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.24.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.24.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.24.ffn_norm.weight
F32
F32
[1536]
blk.24.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.25
blk.25.attn_k.bias
F32
F32
[256]
blk.25.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.25.attn_norm.weight
F32
F32
[1536]
blk.25.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.25.attn_q.bias
F32
F32
[1536]
blk.25.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.25.attn_v.bias
F32
F32
[256]
blk.25.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.25.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.25.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.25.ffn_norm.weight
F32
F32
[1536]
blk.25.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.26
blk.26.attn_k.bias
F32
F32
[256]
blk.26.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.26.attn_norm.weight
F32
F32
[1536]
blk.26.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.26.attn_q.bias
F32
F32
[1536]
blk.26.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.26.attn_v.bias
F32
F32
[256]
blk.26.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.26.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.26.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.26.ffn_norm.weight
F32
F32
[1536]
blk.26.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
blk.27
blk.27.attn_k.bias
F32
F32
[256]
blk.27.attn_k.weight
Q8_0
Q8_0
[1536, 256]
blk.27.attn_norm.weight
F32
F32
[1536]
blk.27.attn_output.weight
Q8_0
Q8_0
[1536, 1536]
blk.27.attn_q.bias
F32
F32
[1536]
blk.27.attn_q.weight
Q8_0
Q8_0
[1536, 1536]
blk.27.attn_v.bias
F32
F32
[256]
blk.27.attn_v.weight
Q8_0
Q8_0
[1536, 256]
blk.27.ffn_down.weight
Q8_0
Q8_0
[8960, 1536]
blk.27.ffn_gate.weight
Q8_0
Q8_0
[1536, 8960]
blk.27.ffn_norm.weight
F32
F32
[1536]
blk.27.ffn_up.weight
Q8_0
Q8_0
[1536, 8960]
output_norm.weight
F32
F32
[1536]