INDEX

Explanations

attends to keywords or identifiers marked with "name" from programming context keywords or operators

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

0:0.06

1:0.13

2:0.10

3:0.03

4:0.17

5:0.37

6:0.04

7:0.06

Negative Logits

ueba

-0.26

Thunk

-0.25

elry

-0.24

ductors

-0.24

zés

-0.23

xca

-0.23

 deut

-0.23

ificance

-0.22

 Micah

-0.22

sient

-0.22

POSITIVE LOGITS

AnimationsModule

0.35

 مشين

0.33

存于互联网档案馆

0.32

PreferredItem

0.32

writeFieldEnd

0.31

ItemBackground

0.31

 ComVisible

0.30

bootstrapcdn

0.30

EndContext

0.28

CloseOperation

0.28

Activations Density 0.125%

attends to keywords or identifiers marked with "name" from programming context keywords or operators

No Comments

No Known Activations