INDEX

Explanations

terms related to fraud and deceit

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 36-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_36/width_16k/average_l0_120

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.36.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

Попис

-0.40

mbangkan

-0.39

 chóng

-0.37

 stanga

-0.37

 hacerlo

-0.36

 зага

-0.36

 übere

-0.35

Handles

-0.35

Únete

-0.35

 controversial

-0.35

POSITIVE LOGITS

 perpetrated

0.57

[]:

0.54

UserScript

0.53

 Prevention

0.52

 defaultstate

0.52

uous

0.50

 tendencies

0.50

 prevention

0.49

الحياه

0.49

 ویکی‌پدیا

0.49

Activations Density 0.609%

terms related to fraud and deceit

No Comments

No Known Activations