INDEX

Explanations

references to the concept of innocence, particularly in the context of people or souls

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

describing innocence

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

 gömlek

-0.55

 réun

-0.49

 vectorielle

-0.49

 stället

-0.48

woordig

-0.47

 elbise

-0.46

 umowy

-0.44

 Spagna

-0.44

 geçir

-0.44

 styr

-0.43

POSITIVE LOGITS

 Innocent

1.05

 innocent

1.05

innocent

1.05

innoc

1.04

Innoc

1.04

 Innocence

0.97

 innocence

0.93

 Innoc

0.93

 innoc

0.90

 Inoc

0.86

Activations Density 0.007%

references to the concept of innocence, particularly in the context of people or souls

describing innocence

No Comments

No Known Activations