INDEX

Explanations

phrases indicating intent or anticipation related to observation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 12-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_12/width_16k/average_l0_130

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.12.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

 Cider

-0.55

ymce

-0.52

째

-0.52

 Mahl

-0.51

casila

-0.50

 deleteById

-0.50

aturing

-0.49

NIA

-0.48

 Henning

-0.48

NFPA

-0.48

POSITIVE LOGITS

 LOOK

1.25

 Look

1.19

Look

1.19

 look

1.18

look

1.18

LOOK

1.09

looking

1.09

 LOOKING

1.05

 looking

1.05

looked

1.05

Activations Density 0.069%

phrases indicating intent or anticipation related to observation

No Comments

No Known Activations