INDEX

Explanations

coded patterns or sequences that represent specific data types or structures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 21-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_21/width_16k/average_l0_70

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.21.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

 pinulongan

-0.88

 تضيفلها

-0.68

__(/*!

-0.67

AxisAlignment

-0.65

")){

-0.63

LikeLike

-0.63

AnimationsModule

-0.62

>");

-0.62

'],

-0.61

)"),

-0.60

POSITIVE LOGITS

 morire

0.68

 flotante

0.64

 Plutarch

0.64

 Cæsar

0.63

 regeringen

0.62

 argint

0.62

 Euphrates

0.61

 viendra

0.60

 umane

0.59

Activations Density 0.166%

coded patterns or sequences that represent specific data types or structures

No Comments

No Known Activations