INDEX

Explanations

titles and roles associated with editorial or directorial positions

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

roles and titles

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

:+:

-0.56

 OMIT

-0.54

 تضيفلها

-0.49

 Thar

-0.48

estilo

-0.46

psum

-0.46

atherapy

-0.46

fortunes

-0.45

 screwed

-0.45

aarrggbb

-0.45

POSITIVE LOGITS

InjectMocks

0.52

 Familienname

0.51

 AssemblyCompany

0.46

Aholisi

0.44

AutoScaleMode

0.43

 dueño

0.42

 responsável

0.40

UseVisualStyle

0.40

 Responsable

0.39

 lecteur

0.39

Activations Density 0.020%

titles and roles associated with editorial or directorial positions

roles and titles

No Comments

No Known Activations