INDEX

Explanations

attends to tokens related to letters or abbreviations from corresponding numbered tokens or markers

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

0:0.56

1:0.17

2:0.07

3:0.04

4:0.04

5:0.02

6:0.03

7:0.04

Negative Logits

Efq

-0.32

JNIEnv

-0.31

 AttributeSet

-0.31

esgue

-0.31

 condamné

-0.30

存于互联网档案馆

-0.29

propTypes

-0.29

RectangleBorder

-0.29

CloseOperation

-0.29

intellij

-0.28

POSITIVE LOGITS

IVEREF

0.37

 estekak

0.36

 للاسماء

0.33

 Walkover

0.30

AndEndTag

0.28

ArrowToggle

0.27

तान

0.26

 الحره

0.26

AddHtmlAttribute

0.26

GOTREF

0.26

Activations Density 0.959%

attends to tokens related to letters or abbreviations from corresponding numbered tokens or markers

No Comments

No Known Activations