aguitachan3/yuuki-3.7:f32/model

aguitachan3/ yuuki-3.7:f32

Updated yesterday

yuuki-3.7:f32 ... /

model

abd94161b3dd · 329MB

Metadata

general.architecture

gpt2

gpt2
general.file_type

F32

F32
gpt2.attention.head_count

12

12
gpt2.attention.layer_norm_epsilon

1e-05

1e-05
gpt2.block_count

6

6
gpt2.context_length

1024

1024
gpt2.embedding_length

768

768
gpt2.feed_forward_length

3072

3072
tokenizer.ggml.bos_token_id

50256

50256
tokenizer.ggml.eos_token_id

50256

50256
tokenizer.ggml.merges

[Ġ t, Ġ a, h e, i n, r e, ...]

[Ġ t, Ġ a, h e, i n, r e, ...]
tokenizer.ggml.model

gpt2

gpt2
tokenizer.ggml.padding_token_id

50256

50256
tokenizer.ggml.pre

gpt-2

gpt-2
tokenizer.ggml.token_type

[1, 1, 1, 1, 1, ...]

[1, 1, 1, 1, 1, ...]
tokenizer.ggml.tokens

[!, ", #, $, %, ...]

[!, ", #, $, %, ...]
tokenizer.ggml.unknown_token_id

50256

50256

Tensor

Name

Type

Shape
token_embd.weight

F32

F32

[768, 50257]

blk.0

blk.0.attn_norm.bias

F32

F32

[768]
blk.0.attn_norm.weight

F32

F32

[768]
blk.0.attn_output.bias

F32

F32

[768]
blk.0.attn_output.weight

F32

F32

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

F32

F32

[2304]
blk.0.attn_qkv.weight

F32

F32

[768, 2304]
blk.0.ffn_down.bias

F32

F32

[768]
blk.0.ffn_down.weight

F32

F32

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

F32

F32

[768]
blk.0.ffn_norm.weight

F32

F32

[768]
blk.0.ffn_up.bias

F32

F32

[3072]
blk.0.ffn_up.weight

F32

F32

[768, 3072]

blk.1

blk.1.attn_norm.bias

F32

F32

[768]
blk.1.attn_norm.weight

F32

F32

[768]
blk.1.attn_output.bias

F32

F32

[768]
blk.1.attn_output.weight

F32

F32

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

F32

F32

[2304]
blk.1.attn_qkv.weight

F32

F32

[768, 2304]
blk.1.ffn_down.bias

F32

F32

[768]
blk.1.ffn_down.weight

F32

F32

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

F32

F32

[768]
blk.1.ffn_norm.weight

F32

F32

[768]
blk.1.ffn_up.bias

F32

F32

[3072]
blk.1.ffn_up.weight

F32

F32

[768, 3072]

blk.2

blk.2.attn_norm.bias

F32

F32

[768]
blk.2.attn_norm.weight

F32

F32

[768]
blk.2.attn_output.bias

F32

F32

[768]
blk.2.attn_output.weight

F32

F32

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

F32

F32

[2304]
blk.2.attn_qkv.weight

F32

F32

[768, 2304]
blk.2.ffn_down.bias

F32

F32

[768]
blk.2.ffn_down.weight

F32

F32

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

F32

F32

[768]
blk.2.ffn_norm.weight

F32

F32

[768]
blk.2.ffn_up.bias

F32

F32

[3072]
blk.2.ffn_up.weight

F32

F32

[768, 3072]

blk.3

blk.3.attn_norm.bias

F32

F32

[768]
blk.3.attn_norm.weight

F32

F32

[768]
blk.3.attn_output.bias

F32

F32

[768]
blk.3.attn_output.weight

F32

F32

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

F32

F32

[2304]
blk.3.attn_qkv.weight

F32

F32

[768, 2304]
blk.3.ffn_down.bias

F32

F32

[768]
blk.3.ffn_down.weight

F32

F32

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

F32

F32

[768]
blk.3.ffn_norm.weight

F32

F32

[768]
blk.3.ffn_up.bias

F32

F32

[3072]
blk.3.ffn_up.weight

F32

F32

[768, 3072]

blk.4

blk.4.attn_norm.bias

F32

F32

[768]
blk.4.attn_norm.weight

F32

F32

[768]
blk.4.attn_output.bias

F32

F32

[768]
blk.4.attn_output.weight

F32

F32

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

F32

F32

[2304]
blk.4.attn_qkv.weight

F32

F32

[768, 2304]
blk.4.ffn_down.bias

F32

F32

[768]
blk.4.ffn_down.weight

F32

F32

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

F32

F32

[768]
blk.4.ffn_norm.weight

F32

F32

[768]
blk.4.ffn_up.bias

F32

F32

[3072]
blk.4.ffn_up.weight

F32

F32

[768, 3072]

blk.5

blk.5.attn_norm.bias

F32

F32

[768]
blk.5.attn_norm.weight

F32

F32

[768]
blk.5.attn_output.bias

F32

F32

[768]
blk.5.attn_output.weight

F32

F32

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

F32

F32

[2304]
blk.5.attn_qkv.weight

F32

F32

[768, 2304]
blk.5.ffn_down.bias

F32

F32

[768]
blk.5.ffn_down.weight

F32

F32

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

F32

F32

[768]
blk.5.ffn_norm.weight

F32

F32

[768]
blk.5.ffn_up.bias

F32

F32

[3072]
blk.5.ffn_up.weight

F32

F32

[768, 3072]

output_norm.bias

F32

F32

[768]
position_embd.weight

F32

F32

[768, 1024]
output_norm.weight

F32

F32

[768]