INDEX

Explanations

words and phrases associated with buying, revising, and questions

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

repeated emphasis

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 31-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_31/width_131k/average_l0_109

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.31.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

rungsseite

-0.66

+#+#

-0.59

AndEndTag

-0.59

enderror

-0.58

SharedDtor

-0.56

RTLI

-0.54

 CURIAM

-0.53

adaptiveStyles

-0.53

setof

-0.52

ngths

-0.52

POSITIVE LOGITS

 everywhere

0.58

 Everywhere

0.47

everywhere

0.42

 everyone

0.37

 baby

0.36

Everywhere

0.36

 dulces

0.36

Obrázky

0.35

 partout

0.35

 good

0.35

Activations Density 0.019%

words and phrases associated with buying, revising, and questions

repeated emphasis

No Comments

No Known Activations