INDEX

Explanations

attends to numerical tokens from various unspecified previous 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

0:0.13

1:0.18

2:0.09

3:0.12

4:0.10

5:0.05

6:0.08

7:0.21

Negative Logits

UseVisualStyle

-0.50

NameInMap

-0.38

mbggenerated

-0.35

AntiForgeryToken

-0.33

aarrggbb

-0.32

addPreferredGap

-0.31

 saites

-0.31

anueva

-0.30

AddTagHelper

-0.30

 héri

-0.30

POSITIVE LOGITS

 estimés

0.37

 חיצוניים

0.30

 glyphicon

0.29

δες

0.28

 BoxFit

0.28

 Krim

0.28

reportWebVitals

0.28

ter

0.28

 Бю

0.28

 trajets

0.27

Activations Density 0.000%

attends to numerical tokens from various unspecified previous tokens

No Comments

No Known Activations