INDEX

Explanations

terms related to prevention and stopping harmful actions or events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 34-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_34/width_16k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.34.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

usermodel

-0.52

 Picchu

-0.50

Either

-0.45

 AssemblyCompany

-0.44

 Absence

-0.44

urlaub

-0.43

Dtor

-0.43

 Савезне

-0.43

 fehl

-0.43

 دیکھیے

-0.42

POSITIVE LOGITS

 unnecessary

0.74

 terjadinya

0.73

 future

0.71

 potential

0.70

 posibles

0.69

 inneces

0.68

 possible

0.68

 further

0.68

 möglichen

0.67

possible

0.66

Activations Density 0.438%

terms related to prevention and stopping harmful actions or events

No Comments

No Known Activations