INDEX

Explanations

proper nouns, especially names and places, within legal or official contexts

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

})*/

-0.51

 delà

-0.47

 AssemblyTitle

-0.44

delwed

-0.44

quele

-0.43

Referenties

-0.43

TheReal

-0.43

paravant

-0.43

ləş

-0.42

 "..\..\..\

-0.41

POSITIVE LOGITS

 CURIAM

0.54

endphp

0.51

 mouseY

0.49

 مشين

0.45

 sweatshirt

0.44

 publisher

0.43

 roof

0.43

 diocese

0.42

modelBuilder

0.42

 visor

0.41

Activations Density 0.051%

proper nouns, especially names and places, within legal or official contexts

No Comments

No Known Activations