INDEX

Explanations

phrases or words related to unsafe situations or actions

oai_token-act-pair · gpt-3.5-turbo

references to safety and the concept of being unsafe

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

orah

-0.85

cence

-0.84

phasis

-0.81

ership

-0.80

braska

-0.80

bernatorial

-0.80

gdala

-0.80

ophers

-0.80

zzo

-0.80

thood

-0.76

POSITIVE LOGITS

 unsafe

1.04

 adolesc

0.78

nesses

0.72

 unle

0.67

NESS

0.67

 hazardous

0.66

ÃįÃį

0.62

 Ukrain

0.62

 unhealthy

0.61

IED

0.60

Activations Density 0.016%

phrases or words related to unsafe situations or actions

references to safety and the concept of being unsafe

No Comments

No Known Activations