INDEX

Explanations

mentions of banned substances or substance abuse-related terms

oai_token-act-pair · gpt-3.5-turbo

references to banned substances and substance abuse

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

DIR

-0.78

rium

-0.67

 Fior

-0.65

 Blazers

-0.65

 Hawks

-0.65

 Slate

-0.64

opp

-0.64

athered

-0.64

 Ellison

-0.63

ros

-0.63

POSITIVE LOGITS

 substance

1.02

 substances

1.00

fulness

0.97

 Substance

0.92

amphetamine

0.90

 abuser

0.84

 ecstasy

0.84

 abusers

0.83

cohol

0.82

abuse

0.81

Activations Density 0.014%

mentions of banned substances or substance abuse-related terms

references to banned substances and substance abuse

No Comments

No Known Activations