INDEX

Explanations

terms related to administrative processes or roles

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

the word "Admin" or "Administrative"

np_acts-logits-general · gemini-2.0-flash

admin or administration

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.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

Yess

-0.56

HTT

-0.53

 Nähr

-0.52

 Martyr

-0.51

 faiths

-0.51

 utafiti

-0.48

__(/*!

-0.47

 epilepsy

-0.47

 VARI

-0.47

TacToe

-0.46

POSITIVE LOGITS

strator

1.04

ISTRATION

1.00

istered

0.98

stration

0.97

 Admin

0.92

 ADMIN

0.89

admins

0.88

administ

0.88

 administrators

0.88

ModelAdmin

0.86

Activations Density 0.048%

terms related to administrative processes or roles

the word "Admin" or "Administrative"

admin or administration

No Comments

No Known Activations