INDEX

Explanations

attends to the rules and regulations referenced by numerical tokens from legal terms and abbreviations

oai_attention-head · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 10-gemmascope-att-16k

Configuration

google/gemma-scope-2b-pt-att/layer_10/width_16k/average_l0_70

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.10.attn.hook_z

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.11

1:0.15

2:0.14

3:0.08

4:0.11

5:0.06

6:0.10

7:0.20

Negative Logits

ことで

-0.24

ē

-0.23

BufferException

-0.23

wen

-0.22

 ве

-0.22

ye

-0.20

nelse

-0.20

{};

-0.20

Geplaatst

-0.20

POSITIVE LOGITS

出版年

0.37

LookAnd

0.35

IsMutable

0.33

addington

0.31

󠁢

0.30

Diweddarwch

0.29

IUrlHelper

0.28

arabe

0.28

ưu

0.28

клопе

0.28

Activations Density 0.186%

attends to the rules and regulations referenced by numerical tokens from legal terms and abbreviations

No Comments

No Known Activations