INDEX

Explanations

attends to numerical tokens from tokens that denote mathematical or statistical operations

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

0:0.43

1:0.11

2:0.10

3:0.06

4:0.07

5:0.08

6:0.03

7:0.06

Negative Logits

 blancs

-0.26

 bianchi

-0.23

 Guinée

-0.23

-0.22

 bibit

-0.22

borderFill

-0.22

IRS

-0.21

 Turquie

-0.21

 feature

-0.21

feature

-0.21

POSITIVE LOGITS

 Paglinawan

0.42

AntiForgeryToken

0.40

StoryboardSegue

0.40

 Normdatei

0.40

}`).

0.38

XmlAccessType

0.38

 defaultstate

0.37

----</

0.37

InjectAttribute

0.36

Personensuche

0.36

Activations Density 1.256%

attends to numerical tokens from tokens that denote mathematical or statistical operations

No Comments

No Known Activations