INDEX

Explanations

references to physical markings or inscriptions

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

 Europeo

-0.37

 europeo

-0.34

 habet

-0.33

 inmigrantes

-0.33

 frères

-0.32

 inimes

-0.32

ฟัง

-0.32

 uppgifter

-0.32

 europeos

-0.32

 soldats

-0.31

POSITIVE LOGITS

 gynhyrchwyd

0.70

 msgs

0.58

 Taktlose

0.56

PostInfinity

0.54

laid

0.54

tagHelperRunner

0.54

mess

0.54

MessageOf

0.52

getMenuInflater

0.52

 Photocase

0.50

Activations Density 0.059%

references to physical markings or inscriptions

No Comments

No Known Activations