INDEX

Explanations

specific proper nouns and names, particularly related to corporations and individuals

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

company and organization names

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

millimeters

-0.36

funeralhome

-0.35

RenderAtEndOf

-0.34

 koup

-0.33

 psicológica

-0.32

 Klicken

-0.32

eqn

-0.31

 BoxDecoration

-0.31

enang

-0.30

 colectiva

-0.30

POSITIVE LOGITS

帖最后由

0.75

 ujednoznacz

0.56

parsedMessage

0.55

Cyfarwyddwr

0.52

MigrationBuilder

0.52

 Савезне

0.49

SpringRunner

0.49

 Wiktionnaire

0.48

 Numerade

0.48

gemeen

0.47

Activations Density 0.228%

specific proper nouns and names, particularly related to corporations and individuals

company and organization names

No Comments

No Known Activations