INDEX

Explanations

references to advertisements and advertising campaigns

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

 fieltro

-0.47

 Oeste

-0.42

 húmedo

-0.36

évaluateur

-0.36

 cálido

-0.35

 orgánico

-0.35

Autoritní

-0.35

řád

-0.35

 prévue

-0.34

 ouverts

-0.34

POSITIVE LOGITS

 sponsors

0.67

RenderAtEndOf

0.67

 commercials

0.66

 companies

0.64

companies

0.63

Aiheesta

0.63

ilever

0.61

 advertisers

0.61

vertisers

0.60

sponsor

0.59

Activations Density 0.023%

references to advertisements and advertising campaigns

No Comments

No Known Activations