INDEX

Explanations

references to non-related or negative contexts in various subjects

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 41-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_41/width_16k/average_l0_113

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.41.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 certainly

-0.41

 most

-0.41

âneo

-0.40

BagLayout

-0.40

 Certainly

-0.38

ValueStyle

-0.38

 unable

-0.38

makeConstraints

-0.37

 anything

-0.37

Certainly

-0.36

POSITIVE LOGITS

violent

0.63

profit

0.62

refundable

0.59

Profit

0.58

 profit

0.58

invasive

0.57

existent

0.57

resident

0.57

alcoholic

0.56

 refundable

0.56

Activations Density 0.258%

references to non-related or negative contexts in various subjects

No Known Activations