INDEX

Explanations

attends to the tokens following the @@ from tokens marked with [[ -]]

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

0:0.25

1:0.19

2:0.13

3:0.07

4:0.05

5:0.10

6:0.08

7:0.09

Negative Logits

LookAnd

-0.36

UnsafeEnabled

-0.32

{}{}

-0.30

bewerken

-0.30

 NSCoder

-0.29

 newBuilder

-0.28

Personendaten

-0.28

://$

-0.26

 createSlice

-0.26

 "..\..\

-0.26

POSITIVE LOGITS

 gynhyrchwyd

0.29

 '+':

0.29

sieu

0.27

ferien

0.27

Üdv

0.27

 تانيه

0.26

luiting

0.26

avoit

0.26

 '/':

0.26

HostException

0.26

Activations Density 0.041%

attends to the tokens following the **@@** from tokens marked with [[ -]]

No Comments

No Known Activations

attends to the tokens following the @@ from tokens marked with [[ -]]