INDEX

Explanations

technical terms or structured data elements related to user permissions and policies

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_33/width_131k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.33.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

MessageOf

-0.68

 <<<<<<<<<<<<<<

-0.47

 ModelExpression

-0.46

Hentet

-0.45

webElementXpaths

-0.44

WebElementEntity

-0.44



-0.43

 ProtoMessage

-0.42

 iNdEx

-0.41

EndGlobalSection

-0.41

POSITIVE LOGITS

 policy

0.69

 Policy

0.67

policy

0.63

 POLICY

0.59

Policy

0.59

 Security

0.59

 security

0.57

 policies

0.56

 granted

0.56

 granting

0.55

Activations Density 0.089%

technical terms or structured data elements related to user permissions and policies

No Comments

No Known Activations