INDEX

Explanations

attends to the token "padding" from numeric tokens that follow it

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

0:0.16

1:0.21

2:0.12

3:0.04

4:0.07

5:0.05

6:0.06

7:0.24

Negative Logits

 مشين

-0.31

fortawesome

-0.30

 Hille

-0.30

 TextAppearance

-0.29

AddHtmlAttribute

-0.29

 intptr

-0.28

AnimationsModule

-0.26

 strlen

-0.26

zzleHttp

-0.26

 <=",

-0.25

POSITIVE LOGITS

IndexPath

0.29

scriptcase

0.27

 stie

0.25

ActionMode

0.24

cosis

0.24

/*

0.24

Према

0.24

MigrationBuilder

0.23

interprétation

0.23

 sandero

0.23

Activations Density 0.034%

attends to the token "padding" from numeric tokens that follow it

No Comments

No Known Activations