INDEX

Explanations

attends to tokens that are abbreviated with specific letters from followed tokens with punctuation

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

0:0.41

1:0.14

2:0.11

3:0.08

4:0.07

5:0.05

6:0.03

7:0.06

Negative Logits

RenderAtEndOf

-0.39

Portale

-0.34

 himself

-0.33

 فريبيس

-0.33

LayoutStyle

-0.32

fillType

-0.32

énario

-0.31

 تضيفلها

-0.30

LookAnd

-0.29

്

-0.29

POSITIVE LOGITS

choenen

0.31

ieteur

0.31

 Bioaccumulative

0.30

enegger

0.29

gister

0.29

 ifrån

0.28

ógy

0.27

 Vlas

0.27

 iParam

0.26

phorbia

0.26

Activations Density 0.619%

attends to tokens that are abbreviated with specific letters from followed tokens with punctuation

No Comments

No Known Activations