INDEX

Explanations

terms related to violence and conflict, particularly in the context of protests and law enforcement actions

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

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

 kasarigan

-0.56

 écrits

-0.45

 utafitiHapana

-0.45

 Commodities

-0.39

 réfugiés

-0.38

issau

-0.37

 surla

-0.37

gidas

-0.36

 literary

-0.36

 tapis

-0.36

POSITIVE LOGITS

 barrica

0.45

 mergeFrom

0.44

 incen

0.43

 injuring

0.43

 photos

0.42

 costumes

0.42

 clip

0.40

 ſte

0.39

 IERC

0.39

 protesters

0.39

Activations Density 0.026%

terms related to violence and conflict, particularly in the context of protests and law enforcement actions

No Comments

No Known Activations