INDEX

Explanations

various forms of the word "bla" and related variations that could indicate dismissive or non-specific content

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

épisode

-0.51

حياته

-0.48

 Italij

-0.47

帖最后由

-0.47

 Cotter

-0.47

styleType

-0.46

pidou

-0.46

 Verhältnisse

-0.45

sprozess

-0.44

quillas

-0.44

POSITIVE LOGITS

BLA

1.15

Bla

1.08

 Blake

1.03

bla

1.00

Bla

0.98

bla

0.96

Blake

0.93

BLA

0.91

 Blaze

0.89

 Blair

0.88

Activations Density 0.008%

various forms of the word "bla" and related variations that could indicate dismissive or non-specific content

No Comments

No Known Activations