INDEX

Explanations

action phrases related to navigating software menus

oai_token-act-pair · gpt-4o-mini Triggered by @bot

and then actions

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

----</

-0.46

fillType

-0.43

Espèce

-0.38

.*")]

-0.38

tagHelperRunner

-0.38

 Qualität

-0.37

twimg

-0.36

Jîn

-0.36

ModelBuilder

-0.35

 criticisms

-0.35

POSITIVE LOGITS

 linkovi

0.56

WebElementEntity

0.50

 sumpay

0.45

 Paglinawan

0.42

Loh

0.41

abb

0.41

XmlRootElement

0.40

ɵɵ

0.40

verifyException

0.40

ⓧ

0.40

Activations Density 0.020%

action phrases related to navigating software menus

and then actions

No Comments

No Known Activations