INDEX

Explanations

expressions of awe and admiration

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

sense of awe or wonder

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.9.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 AssemblyCulture

-0.79

 ModelExpression

-0.73

 Administrativna

-0.68

 desertcart

-0.66

 للاسماء

-0.63

 '\\;'

-0.63

 unknownFields

-0.61

inSlope

-0.60

 Monfieur

-0.60

[@BOS@]

-0.60

POSITIVE LOGITS

awe

0.57

awe

0.39

 beeindruck

0.38

 admire

0.36

PYX

0.34

Awe

0.34

 karş

0.33

 admi

0.33

 impresionantes

0.33

 awesome

0.32

Activations Density 0.156%

expressions of awe and admiration

sense of awe or wonder

No Known Activations