INDEX

Explanations

phrases expressing varying levels of interest or concern regarding topics

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

interest and care

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

 fehler

-0.44

 coraggio

-0.41

 penger

-0.40

makeConstraints

-0.40

Fehler

-0.40

 žal

-0.39

GEBURTS

-0.39

 Medicines

-0.39

 publicitaires

-0.38

 sensação

-0.38

POSITIVE LOGITS

 interest

0.85

 interested

0.77

 interessiert

0.73

Interest

0.73

 Interest

0.71

 interes

0.69

 interés

0.69

 interessieren

0.69

interested

0.67

 Interesse

0.66

Activations Density 0.125%

phrases expressing varying levels of interest or concern regarding topics

interest and care

No Comments

No Known Activations