INDEX

Explanations

instances of manipulation and deceit in context

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

Deception, fraud, or trickery

np_acts-logits-general · gemini-2.0-flash

fraud and theft

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.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

 nakalista

-0.94

MessageTagHelper

-0.84

 initComponents

-0.78

Tracce

-0.76

parsedMessage

-0.75

ValueStyle

-0.74

 архивлан

-0.73

 كومونز

-0.73

الدراسه

-0.70

 Drapeau

-0.69

POSITIVE LOGITS

 trick

0.78

 fooling

0.74

 fooled

0.73

 successfully

0.71

 deceive

0.65

瞞

0.65

trick

0.62

 succeed

0.62

 successful

0.60

 fool

0.60

Activations Density 0.234%

instances of manipulation and deceit in context

Deception, fraud, or trickery

fraud and theft

No Comments

No Known Activations