INDEX

Explanations

phrases indicating social interactions and individual relationships

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

language names and technical terms

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

ium

-0.29

 кӀ

-0.26

 yılında

-0.25

 pernas

-0.25

featureID

-0.25

oredCriteria

-0.24

 удалось

-0.24

 diper

-0.24

новниш

-0.24

Superficie

-0.24

POSITIVE LOGITS

adaptiveStyles

0.65

haikusbot

0.64

новништво

0.63

MLLoader

0.59

httphttps

0.58

iſchen

0.58

niſſe

0.56

iſche

0.55

 Disqus

0.54

 translators

0.53

Activations Density 0.075%

phrases indicating social interactions and individual relationships

language names and technical terms

No Comments

No Known Activations