INDEX

Explanations

geographic or locational references

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

Northern locales and regions

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

parsedMessage

-0.95

featureID

-0.89

Personendaten

-0.88

 sumpay

-0.87

principalColumn

-0.86

 kaarangay

-0.83

 ujednoznacz

-0.81

 Chwiliwch

-0.79

\{\\

-0.77

 noDo

-0.77

POSITIVE LOGITS

 ever

0.28

 among

0.26

 rivals

0.24

0.23

 nejlepší

0.23

の中では

0.21

 lawmakers

0.20

 Amérique

0.20

NE

0.20

 rival

0.20

Activations Density 0.043%

geographic or locational references

Northern locales and regions

No Comments

No Known Activations