INDEX

Explanations

codes or scripts related to data definitions and structures

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

 aikana

-0.42

featureID

-0.40

 quizá

-0.37

 Gebet

-0.36

 reír

-0.35

 zagran

-0.35

 llorando

-0.35

 rocz

-0.35

 obligado

-0.34

 quizás

-0.34

POSITIVE LOGITS

 ***!

0.79

ðsíða

0.75

 Verſ

0.73

 beſte

0.66

 tartalo

0.65

 ſei

0.65

௃

0.63

 zwiſchen

0.63

 vooz

0.63

OGND

0.62

Activations Density 0.060%

codes or scripts related to data definitions and structures

No Comments

No Known Activations