INDEX

Explanations

token markers or structural elements in code or formatted data

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 3-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_3/width_16k/average_l0_90

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.3.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

import

-0.63

enumi

-0.50

 Crime

-0.49

importance

-0.48

 IMPORT

-0.48

Crime

-0.47

crime

-0.46

 Import

-0.46

glVertex

-0.45

Importance

-0.44

POSITIVE LOGITS

},

0.60

tanleria

0.58

ihnachten

0.46

よう

0.46

 CanadaChoose

0.45

誰か

0.44

 actionMode

0.44

})->

0.43

 kaarangay

0.43

</>

0.42

Activations Density 0.124%

token markers or structural elements in code or formatted data

No Comments

No Known Activations