INDEX

Explanations

phrases that indicate secrecy, deception, or social judgment

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 estekak

-0.57

IUrlHelper

-0.54

 EconPapers

-0.49

sizeCache

-0.49

unknownFields

-0.48

XmlAccessType

-0.48

rungsseite

-0.47

 localObject

-0.47

riwal

-0.45



-0.45

POSITIVE LOGITS

 BorderLayout

0.37

PostConstruct

0.34

Économie

0.33

 ragion

0.33

 spoilt

0.32

resave

0.32

peq

0.31

ワーク

0.31

abetes

0.30

Демографія

0.30

Activations Density 0.124%

phrases that indicate secrecy, deception, or social judgment

No Comments

No Known Activations