INDEX

Explanations

geographic locations and names of places

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

place names followed by context

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

 betweenstory

-0.81

 ब्रेकडाउन

-0.79

 queſta

-0.76

<unused28>

-0.71

<unused8>

-0.71

<unused3>

-0.71

<unused47>

-0.71

<unused51>

-0.71

<unused42>

-0.71

<unused23>

-0.71

POSITIVE LOGITS

0.29

lit

0.27

me

0.27

la

0.26

 Coast

0.25

 Mayor

0.25

’

0.24

IContainer

0.23

↵

0.23

Activations Density 0.004%

geographic locations and names of places

place names followed by context

No Comments

No Known Activations