INDEX

Explanations

references to indoor environments or activities

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

indoor spaces and air

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

 violencia

-0.50

évalu

-0.49

hésite

-0.48

owulf

-0.47

 promocional

-0.47

 lehetős

-0.46

 Comunicação

-0.46

 caballos

-0.46

 Gaulle

-0.46

offerta

-0.45

POSITIVE LOGITS

 Indoor

1.07

 indoor

1.06

 indoors

1.03

Indoor

1.03

indoor

0.95

 inside

0.90

Inside

0.88

 Inside

0.83

inside

0.78

 INSIDE

0.75

Activations Density 0.004%

references to indoor environments or activities

indoor spaces and air

No Comments

No Known Activations