INDEX

Explanations

references to possession or personalization

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

 maternidade

-0.45

 spreken

-0.43

 uważ

-0.40

 oleju

-0.40

diensten

-0.39

 kematian

-0.36

 agência

-0.35

 ibunya

-0.35

 publicação

-0.34

 politie

-0.33

POSITIVE LOGITS

 pinulongan

0.66

 Roskov

0.61

 ivelany

0.60

IsContent

0.60

FunctionFlags

0.59

 jsPsych

0.56

argout

0.56

richTextPanel

0.55

 invokingState

0.54

клопе

0.51

Activations Density 0.014%

references to possession or personalization

No Comments

No Known Activations