INDEX

Explanations

attends to percentage tokens from containment-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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.10

1:0.14

2:0.11

3:0.12

4:0.13

5:0.05

6:0.15

7:0.16

Negative Logits

 <>",

-0.36

riors

-0.36

!';

-0.34

*/

-0.33

%%

-0.33

--)

-0.32

tôi

-0.32

_$

-0.31

'>{

-0.31

 surfers

-0.30

POSITIVE LOGITS

tt

0.29

antd

0.29

 intptr

0.28

oa̍t

0.28

 confronti

0.28

 travaillons

0.27

Xaml

0.27

parsedMessage

0.26

databind

0.26

LookAnd

0.26

Activations Density 0.018%

attends to percentage tokens from containment-related tokens

No Comments

No Known Activations