INDEX

Explanations

terms related to deception or misleading tactics

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

Potentially undesirable actions or outcomes

np_acts-logits-general · gemini-2.0-flash

deception and falsehoods

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 22-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_22/width_16k/average_l0_72

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.22.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

TagMode

-0.45

存知

-0.44

 virke

-0.44

RECEIVED

-0.44

didReceive

-0.44

devamını

-0.42

ța

-0.41

şehir

-0.41

จริง

-0.41

lète

-0.41

POSITIVE LOGITS

 ErrIntOverflow

0.96

ftagPool

0.89

 NDEBUG

0.84

ishness

0.81

 Monfieur

0.79

 ſche

0.79

 nonsense

0.76

 galore

0.76

 wireType

0.75

gery

0.75

Activations Density 0.334%

terms related to deception or misleading tactics

Potentially undesirable actions or outcomes

deception and falsehoods

No Comments

No Known Activations