INDEX

Explanations

references to historical locations 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 @ 22-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_22/width_16k/average_l0_123

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.22.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Warszawie

-0.50

 Gauteng

-0.47

 paulista

-0.47

 Lombardia

-0.47

 Selangor

-0.46

 Krakowie

-0.46

 głowy

-0.45

 warszaw

-0.45

 Seattle

-0.44

AnimationsModule

-0.43

POSITIVE LOGITS

 town

0.73

 village

0.55

 utafitiHapana

0.54

Town

0.54

 Town

0.54

 البل

0.53

 villagers

0.48

 TOWN

0.48

town

0.48

 towns

0.48

Activations Density 0.262%

references to historical locations and structures

No Comments

No Known Activations