INDEX

Explanations

mentions of being under the influence of substances

oai_token-act-pair · gpt-3.5-turbo

signals or indicators of drug use and potential intoxication

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

displayText

-0.65

 partName

-0.57

 lineback

-0.56

antage

-0.55

cour

-0.55

igi

-0.53

ctors

-0.52

ofi

-0.51

vironments

-0.50

uitive

-0.50

POSITIVE LOGITS

 inappropriately

0.76

 improperly

0.76

"'

0.72

 nefarious

0.71

 illegally

0.67

 unfairly

0.66

 illegitimate

0.65

 prostitutes

0.64

 improper

0.63

 wrongly

0.63

Activations Density 1.538%

mentions of being under the influence of substances

signals or indicators of drug use and potential intoxication

No Comments

No Known Activations