INDEX

Explanations

phrases and responses indicating agreement or acknowledgment

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

French "merci" and "no"

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

GraphicsUnit

-0.68

XmlAccessType

-0.61

transQ

-0.50

AddTagHelper

-0.49

UrlResolution

-0.47

cerpt

-0.45

hyrchwyd

-0.45

 lapses

-0.45

 تضيفلها

-0.45

'{@

-0.45

POSITIVE LOGITS

setVerticalGroup

0.48

 ISNI

0.39

 uVar

0.36

CloseOperation

0.36

 EClass

0.35

toHaveBeenCalled

0.34

 remercie

0.33

cessite

0.32

 kaarangay

0.31

 conserv

0.31

Activations Density 0.008%

phrases and responses indicating agreement or acknowledgment

French "merci" and "no"

No Comments

No Known Activations