INDEX

Explanations

references to legal and political scandals

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

Autoritní

-0.54

сылкі

-0.39

iet

-0.37

enery

-0.37

SOL

-0.36

 Roskov

-0.36

 autorytatywna

-0.36

ylo

-0.36

 Observing

-0.35

elor

-0.35

POSITIVE LOGITS

 scandals

0.77

 revelations

0.67

 scandal

0.66

 Untersuch

0.64

TypedDataSet

0.60

 investigation

0.59

 investigations

0.59

 crimes

0.59

 allegations

0.54

 investigated

0.53

Activations Density 0.121%

references to legal and political scandals

No Comments

No Known Activations