INDEX

Explanations

comparative and evaluative language related to analysis and definitions

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

contrasting, applicable, defined

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

StructEnd

-0.48

 didalam

-0.45

 dependencia

-0.43

 ambientale

-0.43

 politiche

-0.43

 verdaderas

-0.43

 juist

-0.41

 termica

-0.41

fikation

-0.40

 esferas

-0.40

POSITIVE LOGITS

astify

0.67

setVerticalGroup

0.64

 internetowa

0.58

حوالہ

0.58

حياتها

0.57

RTLR

0.57

GEBURTSDATUM

0.56

المشاركات

0.55

SourceChecksum

0.55

حياته

0.55

Activations Density 0.069%

comparative and evaluative language related to analysis and definitions

contrasting, applicable, defined

No Comments

No Known Activations