INDEX

Explanations

attends to tokens that signify a period or full stop from tokens that denote related topics or items

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

0:0.07

1:0.08

2:0.09

3:0.10

4:0.08

5:0.02

6:0.37

7:0.15

Negative Logits

 autorytatywna

-0.54

 nahilalakip

-0.50

Попис

-0.48

 ISNI

-0.48

Autoritní

-0.47

 AssemblyProduct

-0.46

expandindo

-0.46

 "..\..\..\

-0.45

LookAnd

-0.45

$_['

-0.44

POSITIVE LOGITS

thinkable

0.30

 kiin

0.27

็จ

0.27

xFFFFFF

0.27

GRAPHS

0.27

 lenne

0.26

 ljudi

0.26

 eriş

0.26

inalg

0.25

positiva

0.25

Activations Density 0.001%

attends to tokens that signify a period or full stop from tokens that denote related topics or items

No Comments

No Known Activations