INDEX

Explanations

attends to function-related tokens from related tokens

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

0:0.43

1:0.23

2:0.08

3:0.05

4:0.04

5:0.03

6:0.04

7:0.06

Negative Logits

Erreferentziak

-0.23

 Wilkins

-0.22

 anzu

-0.22

 fréquent

-0.20

upné

-0.20

'&:

-0.20

 geweest

-0.20

 komende

-0.20

unately

-0.20

he

-0.20

POSITIVE LOGITS

IonicModule

0.32

 iſt

0.29

 hObject

0.28

 itſelf

0.27

 useAppContext

0.27

GOTREF

0.26

'{@

0.26

///</

0.25

@"/

0.25

bonucleic

0.24

Activations Density 0.627%

attends to function-related tokens from related tokens

No Comments

No Known Activations