kun432/cl-nagoya-ruri-base/model

kun432/

cl-nagoya-ruri-base:latest

1,478 Downloads Updated 11 months ago

Ruri: Japanese General Text Embeddings

embedding 111m

cl-nagoya-ruri-base:latest ... /

model

4f0427c38bfc · 223MB

Metadata

general.architecture

bert

bert
general.file_type

F16

F16
bert.attention.causal

false

false
bert.attention.head_count

12

12
bert.attention.layer_norm_epsilon

1e-12

1e-12
bert.block_count

12

12
bert.context_length

512

512
bert.embedding_length

768

768
bert.feed_forward_length

3072

3072
bert.pooling_type

Mean

Mean
tokenizer.ggml.model

bert

bert
tokenizer.ggml.padding_token_id

0

0
tokenizer.ggml.pre

ruri-large

ruri-large
tokenizer.ggml.token_type

[3, 3, 3, 3, 3, ...]

[3, 3, 3, 3, 3, ...]
tokenizer.ggml.token_type_count

2

2
tokenizer.ggml.tokens

[[PAD], [UNK], [CLS], [SEP], [MASK], ...]

[[PAD], [UNK], [CLS], [SEP], [MASK], ...]

Tensor

Name

Type

Shape
token_embd.weight

F16

F16

[768, 32768]

blk.0

blk.0.attn_k.bias

F32

F32

[768]
blk.0.attn_k.weight

F16

F16

[768, 768]
blk.0.attn_output.bias

F32

F32

[768]
blk.0.attn_output.weight

F16

F16

[768, 768]
blk.0.attn_output_norm.bias

F32

F32

[768]
blk.0.attn_output_norm.weight

F32

F32

[768]
blk.0.attn_q.bias

F32

F32

[768]
blk.0.attn_q.weight

F16

F16

[768, 768]
blk.0.attn_v.bias

F32

F32

[768]
blk.0.attn_v.weight

F16

F16

[768, 768]
blk.0.ffn_down.bias

F32

F32

[768]
blk.0.ffn_down.weight

F16

F16

[3072, 768]
blk.0.ffn_up.bias

F32

F32

[3072]
blk.0.ffn_up.weight

F16

F16

[768, 3072]
blk.0.layer_output_norm.bias

F32

F32

[768]
blk.0.layer_output_norm.weight

F32

F32

[768]

blk.1

blk.1.attn_k.bias

F32

F32

[768]
blk.1.attn_k.weight

F16

F16

[768, 768]
blk.1.attn_output.bias

F32

F32

[768]
blk.1.attn_output.weight

F16

F16

[768, 768]
blk.1.attn_output_norm.bias

F32

F32

[768]
blk.1.attn_output_norm.weight

F32

F32

[768]
blk.1.attn_q.bias

F32

F32

[768]
blk.1.attn_q.weight

F16

F16

[768, 768]
blk.1.attn_v.bias

F32

F32

[768]
blk.1.attn_v.weight

F16

F16

[768, 768]
blk.1.ffn_down.bias

F32

F32

[768]
blk.1.ffn_down.weight

F16

F16

[3072, 768]
blk.1.ffn_up.bias

F32

F32

[3072]
blk.1.ffn_up.weight

F16

F16

[768, 3072]
blk.1.layer_output_norm.bias

F32

F32

[768]
blk.1.layer_output_norm.weight

F32

F32

[768]

blk.2

blk.2.attn_k.bias

F32

F32

[768]
blk.2.attn_k.weight

F16

F16

[768, 768]
blk.2.attn_output.bias

F32

F32

[768]
blk.2.attn_output.weight

F16

F16

[768, 768]
blk.2.attn_output_norm.bias

F32

F32

[768]
blk.2.attn_output_norm.weight

F32

F32

[768]
blk.2.attn_q.bias

F32

F32

[768]
blk.2.attn_q.weight

F16

F16

[768, 768]
blk.2.attn_v.bias

F32

F32

[768]
blk.2.attn_v.weight

F16

F16

[768, 768]
blk.2.ffn_down.bias

F32

F32

[768]
blk.2.ffn_down.weight

F16

F16

[3072, 768]
blk.2.ffn_up.bias

F32

F32

[3072]
blk.2.ffn_up.weight

F16

F16

[768, 3072]
blk.2.layer_output_norm.bias

F32

F32

[768]
blk.2.layer_output_norm.weight

F32

F32

[768]

blk.3

blk.3.attn_k.bias

F32

F32

[768]
blk.3.attn_k.weight

F16

F16

[768, 768]
blk.3.attn_output.bias

F32

F32

[768]
blk.3.attn_output.weight

F16

F16

[768, 768]
blk.3.attn_output_norm.bias

F32

F32

[768]
blk.3.attn_output_norm.weight

F32

F32

[768]
blk.3.attn_q.bias

F32

F32

[768]
blk.3.attn_q.weight

F16

F16

[768, 768]
blk.3.attn_v.bias

F32

F32

[768]
blk.3.attn_v.weight

F16

F16

[768, 768]
blk.3.ffn_down.bias

F32

F32

[768]
blk.3.ffn_down.weight

F16

F16

[3072, 768]
blk.3.ffn_up.bias

F32

F32

[3072]
blk.3.ffn_up.weight

F16

F16

[768, 3072]
blk.3.layer_output_norm.bias

F32

F32

[768]
blk.3.layer_output_norm.weight

F32

F32

[768]

blk.4

blk.4.attn_k.bias

F32

F32

[768]
blk.4.attn_k.weight

F16

F16

[768, 768]
blk.4.attn_output.bias

F32

F32

[768]
blk.4.attn_output.weight

F16

F16

[768, 768]
blk.4.attn_output_norm.bias

F32

F32

[768]
blk.4.attn_output_norm.weight

F32

F32

[768]
blk.4.attn_q.bias

F32

F32

[768]
blk.4.attn_q.weight

F16

F16

[768, 768]
blk.4.attn_v.bias

F32

F32

[768]
blk.4.attn_v.weight

F16

F16

[768, 768]
blk.4.ffn_down.bias

F32

F32

[768]
blk.4.ffn_down.weight

F16

F16

[3072, 768]
blk.4.ffn_up.bias

F32

F32

[3072]
blk.4.ffn_up.weight

F16

F16

[768, 3072]
blk.4.layer_output_norm.bias

F32

F32

[768]
blk.4.layer_output_norm.weight

F32

F32

[768]

blk.5

blk.5.attn_k.bias

F32

F32

[768]
blk.5.attn_k.weight

F16

F16

[768, 768]
blk.5.attn_output.bias

F32

F32

[768]
blk.5.attn_output.weight

F16

F16

[768, 768]
blk.5.attn_output_norm.bias

F32

F32

[768]
blk.5.attn_output_norm.weight

F32

F32

[768]
blk.5.attn_q.bias

F32

F32

[768]
blk.5.attn_q.weight

F16

F16

[768, 768]
blk.5.attn_v.bias

F32

F32

[768]
blk.5.attn_v.weight

F16

F16

[768, 768]
blk.5.ffn_down.bias

F32

F32

[768]
blk.5.ffn_down.weight

F16

F16

[3072, 768]
blk.5.ffn_up.bias

F32

F32

[3072]
blk.5.ffn_up.weight

F16

F16

[768, 3072]
blk.5.layer_output_norm.bias

F32

F32

[768]
blk.5.layer_output_norm.weight

F32

F32

[768]

blk.6

blk.6.attn_k.bias

F32

F32

[768]
blk.6.attn_k.weight

F16

F16

[768, 768]
blk.6.attn_output.bias

F32

F32

[768]
blk.6.attn_output.weight

F16

F16

[768, 768]
blk.6.attn_output_norm.bias

F32

F32

[768]
blk.6.attn_output_norm.weight

F32

F32

[768]
blk.6.attn_q.bias

F32

F32

[768]
blk.6.attn_q.weight

F16

F16

[768, 768]
blk.6.attn_v.bias

F32

F32

[768]
blk.6.attn_v.weight

F16

F16

[768, 768]
blk.6.ffn_down.bias

F32

F32

[768]
blk.6.ffn_down.weight

F16

F16

[3072, 768]
blk.6.ffn_up.bias

F32

F32

[3072]
blk.6.ffn_up.weight

F16

F16

[768, 3072]
blk.6.layer_output_norm.bias

F32

F32

[768]
blk.6.layer_output_norm.weight

F32

F32

[768]

blk.7

blk.7.attn_k.bias

F32

F32

[768]
blk.7.attn_k.weight

F16

F16

[768, 768]
blk.7.attn_output.bias

F32

F32

[768]
blk.7.attn_output.weight

F16

F16

[768, 768]
blk.7.attn_output_norm.bias

F32

F32

[768]
blk.7.attn_output_norm.weight

F32

F32

[768]
blk.7.attn_q.bias

F32

F32

[768]
blk.7.attn_q.weight

F16

F16

[768, 768]
blk.7.attn_v.bias

F32

F32

[768]
blk.7.attn_v.weight

F16

F16

[768, 768]
blk.7.ffn_down.bias

F32

F32

[768]
blk.7.ffn_down.weight

F16

F16

[3072, 768]
blk.7.ffn_up.bias

F32

F32

[3072]
blk.7.ffn_up.weight

F16

F16

[768, 3072]
blk.7.layer_output_norm.bias

F32

F32

[768]
blk.7.layer_output_norm.weight

F32

F32

[768]

blk.8

blk.8.attn_k.bias

F32

F32

[768]
blk.8.attn_k.weight

F16

F16

[768, 768]
blk.8.attn_output.bias

F32

F32

[768]
blk.8.attn_output.weight

F16

F16

[768, 768]
blk.8.attn_output_norm.bias

F32

F32

[768]
blk.8.attn_output_norm.weight

F32

F32

[768]
blk.8.attn_q.bias

F32

F32

[768]
blk.8.attn_q.weight

F16

F16

[768, 768]
blk.8.attn_v.bias

F32

F32

[768]
blk.8.attn_v.weight

F16

F16

[768, 768]
blk.8.ffn_down.bias

F32

F32

[768]
blk.8.ffn_down.weight

F16

F16

[3072, 768]
blk.8.ffn_up.bias

F32

F32

[3072]
blk.8.ffn_up.weight

F16

F16

[768, 3072]
blk.8.layer_output_norm.bias

F32

F32

[768]
blk.8.layer_output_norm.weight

F32

F32

[768]

blk.9

blk.9.attn_k.bias

F32

F32

[768]
blk.9.attn_k.weight

F16

F16

[768, 768]
blk.9.attn_output.bias

F32

F32

[768]
blk.9.attn_output.weight

F16

F16

[768, 768]
blk.9.attn_output_norm.bias

F32

F32

[768]
blk.9.attn_output_norm.weight

F32

F32

[768]
blk.9.attn_q.bias

F32

F32

[768]
blk.9.attn_q.weight

F16

F16

[768, 768]
blk.9.attn_v.bias

F32

F32

[768]
blk.9.attn_v.weight

F16

F16

[768, 768]
blk.9.ffn_down.bias

F32

F32

[768]
blk.9.ffn_down.weight

F16

F16

[3072, 768]
blk.9.ffn_up.bias

F32

F32

[3072]
blk.9.ffn_up.weight

F16

F16

[768, 3072]
blk.9.layer_output_norm.bias

F32

F32

[768]
blk.9.layer_output_norm.weight

F32

F32

[768]

blk.10

blk.10.attn_k.bias

F32

F32

[768]
blk.10.attn_k.weight

F16

F16

[768, 768]
blk.10.attn_output.bias

F32

F32

[768]
blk.10.attn_output.weight

F16

F16

[768, 768]
blk.10.attn_output_norm.bias

F32

F32

[768]
blk.10.attn_output_norm.weight

F32

F32

[768]
blk.10.attn_q.bias

F32

F32

[768]
blk.10.attn_q.weight

F16

F16

[768, 768]
blk.10.attn_v.bias

F32

F32

[768]
blk.10.attn_v.weight

F16

F16

[768, 768]
blk.10.ffn_down.bias

F32

F32

[768]
blk.10.ffn_down.weight

F16

F16

[3072, 768]
blk.10.ffn_up.bias

F32

F32

[3072]
blk.10.ffn_up.weight

F16

F16

[768, 3072]
blk.10.layer_output_norm.bias

F32

F32

[768]
blk.10.layer_output_norm.weight

F32

F32

[768]

blk.11

blk.11.attn_k.bias

F32

F32

[768]
blk.11.attn_k.weight

F16

F16

[768, 768]
blk.11.attn_output.bias

F32

F32

[768]
blk.11.attn_output.weight

F16

F16

[768, 768]
blk.11.attn_output_norm.bias

F32

F32

[768]
blk.11.attn_output_norm.weight

F32

F32

[768]
blk.11.attn_q.bias

F32

F32

[768]
blk.11.attn_q.weight

F16

F16

[768, 768]
blk.11.attn_v.bias

F32

F32

[768]
blk.11.attn_v.weight

F16

F16

[768, 768]
blk.11.ffn_down.bias

F32

F32

[768]
blk.11.ffn_down.weight

F16

F16

[3072, 768]
blk.11.ffn_up.bias

F32

F32

[3072]
blk.11.ffn_up.weight

F16

F16

[768, 3072]
blk.11.layer_output_norm.bias

F32

F32

[768]
blk.11.layer_output_norm.weight

F32

F32

[768]

position_embd.weight

F32

F32

[768, 512]
token_embd_norm.bias

F32

F32

[768]
token_embd_norm.weight

F32

F32

[768]
token_types.weight

F32

F32

[768, 2]