INDEX

Explanations

mentions of the term "Lord" accompanied by various references to its authority

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_33/width_131k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.33.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

 Flushing

-0.45

 skjø

-0.42

iegel

-0.42

 Emery

-0.42

 Eucalyptus

-0.41

 Woodbury

-0.41

 Mini

-0.40

 Parthen

-0.40

 Parag

-0.39

 Deep

-0.39

POSITIVE LOGITS

 Jefus

0.90

lords

0.62

lord

0.58

 lords

0.55

 ſche

0.54

Története

0.54

 Jesucristo

0.53

Lord

0.52

 PeEnEo

0.52

SBATCH

0.52

Activations Density 0.005%

mentions of the term "Lord" accompanied by various references to its authority

No Comments

No Known Activations