INDEX

Explanations

questions or inquiries involving people and actions

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

who

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

phazard

-0.46

 déta

-0.40

 nejak

-0.37

 complètes

-0.36

 nějak

-0.36

verride

-0.36

只是一

-0.35

🦺

-0.35

 pasada

-0.35

 intentos

-0.35

POSITIVE LOGITS

who

0.81

Who

0.76

Who

0.72

 whom

0.72

who

0.69

rungsseite

0.68

 Administrativna

0.68

kto

0.67

تقاوى

0.66

Quién

0.65

Activations Density 0.032%

questions or inquiries involving people and actions

who

No Comments

No Known Activations