INDEX

Explanations

terms related to protection and safety

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

protecting others or from harm

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

windowFixed

-0.37

ấy

-0.35

 الحره

-0.34

bobox

-0.34

 rendezvous

-0.34

 beards

-0.33

 colorWith

-0.33

 Bored

-0.33

LoadScene

-0.33

ilités

-0.32

POSITIVE LOGITS

 protection

1.34

protection

1.23

 Protection

1.20

 protect

1.20

Protection

1.19

 protecting

1.13

protect

1.12

 PROTECTION

1.12

 protects

1.06

 protección

1.06

Activations Density 0.044%

terms related to protection and safety

protecting others or from harm

No Comments

No Known Activations