INDEX

Explanations

language related to safety and security

oai_token-act-pair · gpt-3.5-turbo

phrases related to safety concerns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

dx

-0.88

issance

-0.85

eric

-0.78

eta

-0.78

igs

-0.75

sth

-0.75

naire

-0.74

sonian

-0.69

yss

-0.66

iguous

-0.64

POSITIVE LOGITS

 safety

1.20

ailability

1.00

safety

0.95

saf

0.87

Þ

0.85

 practition

0.84

 Safety

0.81

Safety

0.80

 condem

0.80

 ingred

0.75

Activations Density 0.021%

language related to safety and security

phrases related to safety concerns

No Comments

No Known Activations