INDEX

Explanations

terms related to protection and preventive measures

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

protection and prevention

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

RuleContext

-0.56

RTEE

-0.52

+:+

-0.49

 NSCoder

-0.45

ніципалі

-0.45

 incomplète

-0.43

lata

-0.43

 OnInit

-0.42

principalColumn

-0.40

[]:

-0.40

POSITIVE LOGITS

Protect

0.62

 protect

0.60

 safeguard

0.58

protect

0.55

 Protect

0.54

 protector

0.53

 protec

0.51

Protecting

0.50

 defend

0.50

 protects

0.50

Activations Density 0.107%

terms related to protection and preventive measures

protection and prevention

No Comments

No Known Activations