lstep / nxcode-32k

70 Pulls Updated 5 months ago

1 Tag

nxcode-32k:q8_0 ... /

model

51738369de75 · 7.7GB

Metadata

general.architecture

qwen2
general.file_type

7
general.name

Nxcode-CQ-7B-orpo
general.quantization_version

2
quantize.imatrix.chunks_count

98
quantize.imatrix.dataset

/training_data/groups_merged.txt
quantize.imatrix.entries_count

224
quantize.imatrix.file

/models/Nxcode-CQ-7B-orpo-GGUF/Nxcode-CQ-7B-orpo.imatrix
qwen2.attention.head_count

32
qwen2.attention.head_count_kv

4
qwen2.attention.layer_norm_rms_epsilon

1e-05
qwen2.block_count

32
qwen2.context_length

65536
qwen2.embedding_length

4096
qwen2.feed_forward_length

13440
qwen2.rope.freq_base

1e+06
tokenizer.ggml.add_bos_token

false
tokenizer.ggml.add_eos_token

false
tokenizer.ggml.bos_token_id

2
tokenizer.ggml.eos_token_id

4
tokenizer.ggml.model

llama
tokenizer.ggml.padding_token_id

92298
tokenizer.ggml.pre

default
tokenizer.ggml.scores

[0, 0, 0, 0, 0, ...]
tokenizer.ggml.token_type

[2, 3, 3, 3, 3, ...]
tokenizer.ggml.tokens

[<unk>, <s>, <|endoftext|>, <|im_start|>, <|im_end|>, ...]
tokenizer.ggml.unknown_token_id

0

Tensor

Name

Type

Shape
token_embd.weight

Q8_0

[4096, 92416]

blk.0

blk.0.attn_k.bias

F32

[512]
blk.0.attn_k.weight

Q8_0

[4096, 512]
blk.0.attn_norm.weight

F32

[4096]
blk.0.attn_output.weight

Q8_0

[4096, 4096]
blk.0.attn_q.bias

F32

[4096]
blk.0.attn_q.weight

Q8_0

[4096, 4096]
blk.0.attn_v.bias

F32

[512]
blk.0.attn_v.weight

Q8_0

[4096, 512]
blk.0.ffn_down.weight

Q8_0

[13440, 4096]
blk.0.ffn_gate.weight

Q8_0

[4096, 13440]
blk.0.ffn_norm.weight

F32

[4096]
blk.0.ffn_up.weight

Q8_0

[4096, 13440]

blk.1

blk.1.attn_k.bias

F32

[512]
blk.1.attn_k.weight

Q8_0

[4096, 512]
blk.1.attn_norm.weight

F32

[4096]
blk.1.attn_output.weight

Q8_0

[4096, 4096]
blk.1.attn_q.bias

F32

[4096]
blk.1.attn_q.weight

Q8_0

[4096, 4096]
blk.1.attn_v.bias

F32

[512]
blk.1.attn_v.weight

Q8_0

[4096, 512]
blk.1.ffn_down.weight

Q8_0

[13440, 4096]
blk.1.ffn_gate.weight

Q8_0

[4096, 13440]
blk.1.ffn_norm.weight

F32

[4096]
blk.1.ffn_up.weight

Q8_0

[4096, 13440]

blk.2

blk.2.attn_k.bias

F32

[512]
blk.2.attn_k.weight

Q8_0

[4096, 512]
blk.2.attn_norm.weight

F32

[4096]
blk.2.attn_output.weight

Q8_0

[4096, 4096]
blk.2.attn_q.bias

F32

[4096]
blk.2.attn_q.weight

Q8_0

[4096, 4096]
blk.2.attn_v.bias

F32

[512]
blk.2.attn_v.weight

Q8_0

[4096, 512]
blk.2.ffn_down.weight

Q8_0

[13440, 4096]
blk.2.ffn_gate.weight

Q8_0

[4096, 13440]
blk.2.ffn_norm.weight

F32

[4096]
blk.2.ffn_up.weight

Q8_0

[4096, 13440]

blk.3

blk.3.attn_k.bias

F32

[512]
blk.3.attn_k.weight

Q8_0

[4096, 512]
blk.3.attn_norm.weight

F32

[4096]
blk.3.attn_output.weight

Q8_0

[4096, 4096]
blk.3.attn_q.bias

F32

[4096]
blk.3.attn_q.weight

Q8_0

[4096, 4096]
blk.3.attn_v.bias

F32

[512]
blk.3.attn_v.weight

Q8_0

[4096, 512]
blk.3.ffn_down.weight

Q8_0

[13440, 4096]
blk.3.ffn_gate.weight

Q8_0

[4096, 13440]
blk.3.ffn_norm.weight

F32

[4096]
blk.3.ffn_up.weight

Q8_0

[4096, 13440]

blk.4

blk.4.attn_k.bias

F32

[512]
blk.4.attn_k.weight

Q8_0

[4096, 512]
blk.4.attn_norm.weight

F32

[4096]
blk.4.attn_output.weight

Q8_0

[4096, 4096]
blk.4.attn_q.bias

F32

[4096]
blk.4.attn_q.weight

Q8_0

[4096, 4096]
blk.4.attn_v.bias

F32

[512]
blk.4.attn_v.weight

Q8_0

[4096, 512]
blk.4.ffn_down.weight

Q8_0

[13440, 4096]
blk.4.ffn_gate.weight

Q8_0

[4096, 13440]
blk.4.ffn_norm.weight

F32

[4096]
blk.4.ffn_up.weight

Q8_0

[4096, 13440]

blk.5

blk.5.attn_k.bias

F32

[512]
blk.5.attn_k.weight

Q8_0

[4096, 512]
blk.5.attn_norm.weight

F32

[4096]
blk.5.attn_output.weight

Q8_0

[4096, 4096]
blk.5.attn_q.bias

F32

[4096]
blk.5.attn_q.weight

Q8_0

[4096, 4096]
blk.5.attn_v.bias

F32

[512]
blk.5.attn_v.weight

Q8_0

[4096, 512]
blk.5.ffn_down.weight

Q8_0

[13440, 4096]
blk.5.ffn_gate.weight

Q8_0

[4096, 13440]
blk.5.ffn_norm.weight

F32

[4096]
blk.5.ffn_up.weight

Q8_0

[4096, 13440]

blk.6

blk.6.attn_k.bias

F32

[512]
blk.6.attn_k.weight

Q8_0

[4096, 512]
blk.6.attn_norm.weight

F32

[4096]
blk.6.attn_output.weight

Q8_0

[4096, 4096]
blk.6.attn_q.bias

F32

[4096]
blk.6.attn_q.weight

Q8_0

[4096, 4096]
blk.6.attn_v.bias

F32

[512]
blk.6.attn_v.weight

Q8_0

[4096, 512]
blk.6.ffn_down.weight

Q8_0

[13440, 4096]
blk.6.ffn_gate.weight

Q8_0

[4096, 13440]
blk.6.ffn_norm.weight

F32

[4096]
blk.6.ffn_up.weight

Q8_0

[4096, 13440]

blk.7

blk.7.attn_k.bias

F32

[512]
blk.7.attn_k.weight

Q8_0

[4096, 512]
blk.7.attn_norm.weight

F32

[4096]
blk.7.attn_output.weight

Q8_0

[4096, 4096]
blk.7.attn_q.bias

F32

[4096]
blk.7.attn_q.weight

Q8_0

[4096, 4096]
blk.7.attn_v.bias

F32

[512]
blk.7.attn_v.weight

Q8_0

[4096, 512]
blk.7.ffn_down.weight

Q8_0

[13440, 4096]
blk.7.ffn_gate.weight

Q8_0

[4096, 13440]
blk.7.ffn_norm.weight

F32

[4096]
blk.7.ffn_up.weight

Q8_0

[4096, 13440]

blk.8

blk.8.attn_k.bias

F32

[512]
blk.8.attn_k.weight

Q8_0

[4096, 512]
blk.8.attn_norm.weight

F32

[4096]
blk.8.attn_output.weight

Q8_0

[4096, 4096]
blk.8.attn_q.bias

F32

[4096]
blk.8.attn_q.weight

Q8_0

[4096, 4096]
blk.8.attn_v.bias

F32

[512]
blk.8.attn_v.weight

Q8_0

[4096, 512]
blk.8.ffn_down.weight

Q8_0

[13440, 4096]
blk.8.ffn_gate.weight

Q8_0

[4096, 13440]
blk.8.ffn_norm.weight

F32

[4096]
blk.8.ffn_up.weight

Q8_0

[4096, 13440]

blk.9

blk.9.attn_k.bias

F32

[512]
blk.9.attn_k.weight

Q8_0

[4096, 512]
blk.9.attn_norm.weight

F32

[4096]
blk.9.attn_output.weight

Q8_0

[4096, 4096]
blk.9.attn_q.bias

F32

[4096]
blk.9.attn_q.weight

Q8_0

[4096, 4096]
blk.9.attn_v.bias

F32

[512]
blk.9.attn_v.weight

Q8_0

[4096, 512]
blk.9.ffn_down.weight

Q8_0

[13440, 4096]
blk.9.ffn_gate.weight

Q8_0

[4096, 13440]
blk.9.ffn_norm.weight

F32

[4096]
blk.9.ffn_up.weight

Q8_0

[4096, 13440]

blk.10

blk.10.attn_k.bias

F32

[512]
blk.10.attn_k.weight

Q8_0

[4096, 512]
blk.10.attn_norm.weight

F32

[4096]
blk.10.attn_output.weight

Q8_0

[4096, 4096]
blk.10.attn_q.bias

F32

[4096]
blk.10.attn_q.weight

Q8_0

[4096, 4096]
blk.10.attn_v.bias

F32

[512]
blk.10.attn_v.weight

Q8_0

[4096, 512]
blk.10.ffn_down.weight

Q8_0

[13440, 4096]
blk.10.ffn_gate.weight

Q8_0

[4096, 13440]
blk.10.ffn_norm.weight

F32

[4096]
blk.10.ffn_up.weight

Q8_0

[4096, 13440]

blk.11

blk.11.attn_k.bias

F32

[512]
blk.11.attn_k.weight

Q8_0

[4096, 512]
blk.11.attn_norm.weight

F32

[4096]
blk.11.attn_output.weight

Q8_0

[4096, 4096]
blk.11.attn_q.bias

F32

[4096]
blk.11.attn_q.weight

Q8_0

[4096, 4096]
blk.11.attn_v.bias

F32

[512]
blk.11.attn_v.weight

Q8_0

[4096, 512]
blk.11.ffn_down.weight

Q8_0

[13440, 4096]
blk.11.ffn_gate.weight

Q8_0

[4096, 13440]
blk.11.ffn_norm.weight

F32

[4096]
blk.11.ffn_up.weight

Q8_0

[4096, 13440]

blk.12

blk.12.attn_k.bias

F32

[512]
blk.12.attn_k.weight

Q8_0

[4096, 512]
blk.12.attn_norm.weight

F32

[4096]
blk.12.attn_output.weight

Q8_0

[4096, 4096]
blk.12.attn_q.bias

F32

[4096]
blk.12.attn_q.weight

Q8_0

[4096, 4096]
blk.12.attn_v.bias

F32

[512]
blk.12.attn_v.weight

Q8_0

[4096, 512]
blk.12.ffn_down.weight

Q8_0

[13440, 4096]
blk.12.ffn_gate.weight

Q8_0

[4096, 13440]
blk.12.ffn_norm.weight

F32

[4096]
blk.12.ffn_up.weight

Q8_0

[4096, 13440]

blk.13

blk.13.attn_k.bias

F32

[512]
blk.13.attn_k.weight

Q8_0

[4096, 512]
blk.13.attn_norm.weight

F32

[4096]
blk.13.attn_output.weight

Q8_0

[4096, 4096]
blk.13.attn_q.bias

F32

[4096]
blk.13.attn_q.weight

Q8_0

[4096, 4096]
blk.13.attn_v.bias

F32

[512]
blk.13.attn_v.weight

Q8_0

[4096, 512]
blk.13.ffn_down.weight

Q8_0

[13440, 4096]
blk.13.ffn_gate.weight

Q8_0

[4096, 13440]
blk.13.ffn_norm.weight

F32

[4096]
blk.13.ffn_up.weight

Q8_0

[4096, 13440]

blk.14

blk.14.attn_k.bias

F32

[512]
blk.14.attn_k.weight

Q8_0

[4096, 512]
blk.14.attn_norm.weight

F32

[4096]
blk.14.attn_output.weight

Q8_0

[4096, 4096]
blk.14.attn_q.bias

F32

[4096]
blk.14.attn_q.weight

Q8_0

[4096, 4096]
blk.14.attn_v.bias

F32

[512]
blk.14.attn_v.weight

Q8_0

[4096, 512]
blk.14.ffn_down.weight

Q8_0

[13440, 4096]
blk.14.ffn_gate.weight

Q8_0

[4096, 13440]
blk.14.ffn_norm.weight

F32

[4096]
blk.14.ffn_up.weight

Q8_0

[4096, 13440]

blk.15

blk.15.attn_k.bias

F32

[512]
blk.15.attn_k.weight

Q8_0

[4096, 512]
blk.15.attn_norm.weight

F32

[4096]
blk.15.attn_output.weight

Q8_0

[4096, 4096]
blk.15.attn_q.bias

F32

[4096]
blk.15.attn_q.weight

Q8_0

[4096, 4096]
blk.15.attn_v.bias

F32

[512]
blk.15.attn_v.weight

Q8_0

[4096, 512]
blk.15.ffn_down.weight

Q8_0

[13440, 4096]
blk.15.ffn_gate.weight

Q8_0

[4096, 13440]
blk.15.ffn_norm.weight

F32

[4096]
blk.15.ffn_up.weight

Q8_0

[4096, 13440]

blk.16

blk.16.attn_k.bias

F32

[512]
blk.16.attn_k.weight

Q8_0

[4096, 512]
blk.16.attn_norm.weight

F32

[4096]
blk.16.attn_output.weight

Q8_0

[4096, 4096]
blk.16.attn_q.bias

F32

[4096]
blk.16.attn_q.weight

Q8_0

[4096, 4096]
blk.16.attn_v.bias

F32

[512]
blk.16.attn_v.weight

Q8_0

[4096, 512]
blk.16.ffn_down.weight

Q8_0

[13440, 4096]
blk.16.ffn_gate.weight

Q8_0

[4096, 13440]
blk.16.ffn_norm.weight

F32

[4096]
blk.16.ffn_up.weight

Q8_0

[4096, 13440]

blk.17

blk.17.attn_k.bias

F32

[512]
blk.17.attn_k.weight

Q8_0

[4096, 512]
blk.17.attn_norm.weight

F32

[4096]
blk.17.attn_output.weight

Q8_0

[4096, 4096]
blk.17.attn_q.bias

F32

[4096]
blk.17.attn_q.weight

Q8_0

[4096, 4096]
blk.17.attn_v.bias

F32

[512]
blk.17.attn_v.weight

Q8_0

[4096, 512]
blk.17.ffn_down.weight

Q8_0

[13440, 4096]
blk.17.ffn_gate.weight

Q8_0

[4096, 13440]
blk.17.ffn_norm.weight

F32

[4096]
blk.17.ffn_up.weight

Q8_0

[4096, 13440]

blk.18

blk.18.attn_k.bias

F32

[512]
blk.18.attn_k.weight

Q8_0

[4096, 512]
blk.18.attn_norm.weight

F32

[4096]
blk.18.attn_output.weight

Q8_0

[4096, 4096]
blk.18.attn_q.bias

F32

[4096]
blk.18.attn_q.weight

Q8_0

[4096, 4096]
blk.18.attn_v.bias

F32

[512]
blk.18.attn_v.weight

Q8_0

[4096, 512]
blk.18.ffn_down.weight

Q8_0

[13440, 4096]
blk.18.ffn_gate.weight

Q8_0

[4096, 13440]
blk.18.ffn_norm.weight

F32

[4096]
blk.18.ffn_up.weight

Q8_0

[4096, 13440]

blk.19

blk.19.attn_k.bias

F32

[512]
blk.19.attn_k.weight

Q8_0

[4096, 512]
blk.19.attn_norm.weight

F32

[4096]
blk.19.attn_output.weight

Q8_0

[4096, 4096]
blk.19.attn_q.bias

F32

[4096]
blk.19.attn_q.weight

Q8_0

[4096, 4096]
blk.19.attn_v.bias

F32

[512]
blk.19.attn_v.weight

Q8_0

[4096, 512]
blk.19.ffn_down.weight

Q8_0

[13440, 4096]
blk.19.ffn_gate.weight

Q8_0

[4096, 13440]
blk.19.ffn_norm.weight

F32

[4096]
blk.19.ffn_up.weight

Q8_0

[4096, 13440]

blk.20

blk.20.attn_k.bias

F32

[512]
blk.20.attn_k.weight

Q8_0

[4096, 512]
blk.20.attn_norm.weight

F32

[4096]
blk.20.attn_output.weight

Q8_0

[4096, 4096]
blk.20.attn_q.bias

F32

[4096]
blk.20.attn_q.weight

Q8_0

[4096, 4096]
blk.20.attn_v.bias

F32

[512]
blk.20.attn_v.weight

Q8_0

[4096, 512]
blk.20.ffn_down.weight

Q8_0

[13440, 4096]
blk.20.ffn_gate.weight

Q8_0

[4096, 13440]
blk.20.ffn_norm.weight

F32

[4096]
blk.20.ffn_up.weight

Q8_0

[4096, 13440]

blk.21

blk.21.attn_k.bias

F32

[512]
blk.21.attn_k.weight

Q8_0

[4096, 512]
blk.21.attn_norm.weight

F32

[4096]
blk.21.attn_output.weight

Q8_0

[4096, 4096]
blk.21.attn_q.bias

F32

[4096]
blk.21.attn_q.weight

Q8_0

[4096, 4096]
blk.21.attn_v.bias

F32

[512]
blk.21.attn_v.weight

Q8_0

[4096, 512]
blk.21.ffn_down.weight

Q8_0

[13440, 4096]
blk.21.ffn_gate.weight

Q8_0

[4096, 13440]
blk.21.ffn_norm.weight

F32

[4096]
blk.21.ffn_up.weight

Q8_0

[4096, 13440]

blk.22

blk.22.attn_k.bias

F32

[512]
blk.22.attn_k.weight

Q8_0

[4096, 512]
blk.22.attn_norm.weight

F32

[4096]
blk.22.attn_output.weight

Q8_0

[4096, 4096]
blk.22.attn_q.bias

F32

[4096]
blk.22.attn_q.weight

Q8_0

[4096, 4096]
blk.22.attn_v.bias

F32

[512]
blk.22.attn_v.weight

Q8_0

[4096, 512]
blk.22.ffn_down.weight

Q8_0

[13440, 4096]
blk.22.ffn_gate.weight

Q8_0

[4096, 13440]
blk.22.ffn_norm.weight

F32

[4096]
blk.22.ffn_up.weight

Q8_0

[4096, 13440]

blk.23

blk.23.attn_k.bias

F32

[512]
blk.23.attn_k.weight

Q8_0

[4096, 512]
blk.23.attn_norm.weight

F32

[4096]
blk.23.attn_output.weight

Q8_0

[4096, 4096]
blk.23.attn_q.bias

F32

[4096]
blk.23.attn_q.weight

Q8_0

[4096, 4096]
blk.23.attn_v.bias

F32

[512]
blk.23.attn_v.weight

Q8_0

[4096, 512]
blk.23.ffn_down.weight

Q8_0

[13440, 4096]
blk.23.ffn_gate.weight

Q8_0

[4096, 13440]
blk.23.ffn_norm.weight

F32

[4096]
blk.23.ffn_up.weight

Q8_0

[4096, 13440]

blk.24

blk.24.attn_k.bias

F32

[512]
blk.24.attn_k.weight

Q8_0

[4096, 512]
blk.24.attn_norm.weight

F32

[4096]
blk.24.attn_output.weight

Q8_0

[4096, 4096]
blk.24.attn_q.bias

F32

[4096]
blk.24.attn_q.weight

Q8_0

[4096, 4096]
blk.24.attn_v.bias

F32

[512]
blk.24.attn_v.weight

Q8_0

[4096, 512]
blk.24.ffn_down.weight

Q8_0

[13440, 4096]
blk.24.ffn_gate.weight

Q8_0

[4096, 13440]
blk.24.ffn_norm.weight

F32

[4096]
blk.24.ffn_up.weight

Q8_0

[4096, 13440]

blk.25

blk.25.attn_k.bias

F32

[512]
blk.25.attn_k.weight

Q8_0

[4096, 512]
blk.25.attn_norm.weight

F32

[4096]
blk.25.attn_output.weight

Q8_0

[4096, 4096]
blk.25.attn_q.bias

F32

[4096]
blk.25.attn_q.weight

Q8_0

[4096, 4096]
blk.25.attn_v.bias

F32

[512]
blk.25.attn_v.weight

Q8_0

[4096, 512]
blk.25.ffn_down.weight

Q8_0

[13440, 4096]
blk.25.ffn_gate.weight

Q8_0

[4096, 13440]
blk.25.ffn_norm.weight

F32

[4096]
blk.25.ffn_up.weight

Q8_0

[4096, 13440]

blk.26

blk.26.attn_k.bias

F32

[512]
blk.26.attn_k.weight

Q8_0

[4096, 512]
blk.26.attn_norm.weight

F32

[4096]
blk.26.attn_output.weight

Q8_0

[4096, 4096]
blk.26.attn_q.bias

F32

[4096]
blk.26.attn_q.weight

Q8_0

[4096, 4096]
blk.26.attn_v.bias

F32

[512]
blk.26.attn_v.weight

Q8_0

[4096, 512]
blk.26.ffn_down.weight

Q8_0

[13440, 4096]
blk.26.ffn_gate.weight

Q8_0

[4096, 13440]
blk.26.ffn_norm.weight

F32

[4096]
blk.26.ffn_up.weight

Q8_0

[4096, 13440]

blk.27

blk.27.attn_k.bias

F32

[512]
blk.27.attn_k.weight

Q8_0

[4096, 512]
blk.27.attn_norm.weight

F32

[4096]
blk.27.attn_output.weight

Q8_0

[4096, 4096]
blk.27.attn_q.bias

F32

[4096]
blk.27.attn_q.weight

Q8_0

[4096, 4096]
blk.27.attn_v.bias

F32

[512]
blk.27.attn_v.weight

Q8_0

[4096, 512]
blk.27.ffn_down.weight

Q8_0

[13440, 4096]
blk.27.ffn_gate.weight

Q8_0

[4096, 13440]
blk.27.ffn_norm.weight

F32

[4096]
blk.27.ffn_up.weight

Q8_0

[4096, 13440]

blk.28

blk.28.attn_k.bias

F32

[512]
blk.28.attn_k.weight

Q8_0

[4096, 512]
blk.28.attn_norm.weight

F32

[4096]
blk.28.attn_output.weight

Q8_0

[4096, 4096]
blk.28.attn_q.bias

F32

[4096]
blk.28.attn_q.weight

Q8_0

[4096, 4096]
blk.28.attn_v.bias

F32

[512]
blk.28.attn_v.weight

Q8_0

[4096, 512]
blk.28.ffn_down.weight

Q8_0

[13440, 4096]
blk.28.ffn_gate.weight

Q8_0

[4096, 13440]
blk.28.ffn_norm.weight

F32

[4096]
blk.28.ffn_up.weight

Q8_0

[4096, 13440]

blk.29

blk.29.attn_k.bias

F32

[512]
blk.29.attn_k.weight

Q8_0

[4096, 512]
blk.29.attn_norm.weight

F32

[4096]
blk.29.attn_output.weight

Q8_0

[4096, 4096]
blk.29.attn_q.bias

F32

[4096]
blk.29.attn_q.weight

Q8_0

[4096, 4096]
blk.29.attn_v.bias

F32

[512]
blk.29.attn_v.weight

Q8_0

[4096, 512]
blk.29.ffn_down.weight

Q8_0

[13440, 4096]
blk.29.ffn_gate.weight

Q8_0

[4096, 13440]
blk.29.ffn_norm.weight

F32

[4096]
blk.29.ffn_up.weight

Q8_0

[4096, 13440]

blk.30

blk.30.attn_k.bias

F32

[512]
blk.30.attn_k.weight

Q8_0

[4096, 512]
blk.30.attn_norm.weight

F32

[4096]
blk.30.attn_output.weight

Q8_0

[4096, 4096]
blk.30.attn_q.bias

F32

[4096]
blk.30.attn_q.weight

Q8_0

[4096, 4096]
blk.30.attn_v.bias

F32

[512]
blk.30.attn_v.weight

Q8_0

[4096, 512]
blk.30.ffn_down.weight

Q8_0

[13440, 4096]
blk.30.ffn_gate.weight

Q8_0

[4096, 13440]
blk.30.ffn_norm.weight

F32

[4096]
blk.30.ffn_up.weight

Q8_0

[4096, 13440]

blk.31

blk.31.attn_k.bias

F32

[512]
blk.31.attn_k.weight

Q8_0

[4096, 512]
blk.31.attn_norm.weight

F32

[4096]
blk.31.attn_output.weight

Q8_0

[4096, 4096]
blk.31.attn_q.bias

F32

[4096]
blk.31.attn_q.weight

Q8_0

[4096, 4096]
blk.31.attn_v.bias

F32

[512]
blk.31.attn_v.weight

Q8_0

[4096, 512]
blk.31.ffn_down.weight

Q8_0

[13440, 4096]
blk.31.ffn_gate.weight

Q8_0

[4096, 13440]
blk.31.ffn_norm.weight

F32

[4096]
blk.31.ffn_up.weight

Q8_0

[4096, 13440]

output.weight

Q8_0

[4096, 92416]
output_norm.weight

F32

[4096]