INDEX

Explanations

attends to potential-related tokens from subsequent tokens specifying their context or application

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

0:0.17

1:0.16

2:0.39

3:0.05

4:0.03

5:0.02

6:0.03

7:0.11

Negative Logits

 vecteur

-0.28

 '\\;'

-0.28

MigrationBuilder

-0.28

دانشنامهٔ

-0.26

loß

-0.25

carol

-0.25

 resourceCulture

-0.24

ⓧ

-0.24

 encuentros

-0.24

 oportunidad

-0.24

POSITIVE LOGITS

//

0.38

endphp

0.32

脚注の使い方

0.30

Datuak

0.30

enumi

0.29

0.28

 ویکی‌پدیای

0.28

stdc

0.28

 myſelf

0.28

مصادر

0.28

Activations Density 0.497%

attends to potential-related tokens from subsequent tokens specifying their context or application

No Comments

No Known Activations