INDEX

Explanations

concepts related to safety and risk management

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

safety and hazard

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.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

 الحره

-0.66

 kasarigan

-0.62

ConstraintMaker

-0.60

 înc

-0.57

 Geſch

-0.57

windowFixed

-0.56

 Normdatei

-0.56

𓃵

-0.55

 matéria

-0.55

 esprits

-0.54

POSITIVE LOGITS

 safety

0.64

 Safety

0.54

 SAFETY

0.52

Safety

0.52

 unsafe

0.50

 dangerous

0.47

safety

0.44

SAFETY

0.44

 safer

0.42

 emergency

0.41

Activations Density 0.013%

concepts related to safety and risk management

safety and hazard

No Comments

No Known Activations