INDEX

Explanations

references to legal warrants

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

Goodbye

-0.48

 Finley

-0.46

 Fili

-0.44

 Layla

-0.44

Hi

-0.44

 BABY

-0.43

ifun

-0.43

 Frü

-0.43

Frozen

-0.42

Loser

-0.42

POSITIVE LOGITS

 Warrant

1.05

warrant

0.96

 warrant

0.93

 WARRANT

0.88

arrant

0.79

 warrants

0.72

 warranted

0.69

WARR

0.69

 ब्रेकडाउन

0.61

 warran

0.59

Activations Density 0.007%

references to legal warrants

No Comments

No Known Activations