INDEX

Explanations

phrases about protection or safeguarding from harm

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

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

 calcetines

-0.43

 chrétien

-0.41

 stället

-0.39

 hospitales

-0.38

 AssemblyCulture

-0.38

 föres

-0.37

 ilmo

-0.36

 uppgifter

-0.36

 libremente

-0.35

 mères

-0.35

POSITIVE LOGITS

 pinulongan

0.73

CppMethod

0.65

inSlope

0.63

IsMutable

0.58

原始内容存档于

0.58

 unwanted

0.57

 avoid

0.56

 EconPapers

0.55

 beginnetje

0.55

extAlignment

0.55

Activations Density 0.022%

phrases about protection or safeguarding from harm

No Comments

No Known Activations