INDEX

Explanations

warning messages or alerts

oai_token-act-pair · gpt-3.5-turbo

warning statements related to potential dangers or sensitive content

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

aepernick

-0.85

Laughs

-0.78

Favorite

-0.76

ichick

-0.76

obbies

-0.76

Interview

-0.74

brates

-0.73

hement

-0.71

chens

-0.69

 excuse

-0.69

POSITIVE LOGITS

 dangers

1.45

 impending

1.24

 risks

1.18

 danger

1.16

 pitfalls

1.12

 beware

1.06

 dire

1.05

 imminent

1.03

danger

0.96

 consequences

0.95

Activations Density 0.268%

warning messages or alerts

warning statements related to potential dangers or sensitive content

No Comments

No Known Activations