INDEX

Explanations

words related to inappropriate behavior or language

oai_token-act-pair · gpt-3.5-turbo

terms related to indecency and obscenity

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

olin

-0.81

quickShipAvailable

-0.79

ACP

-0.77

iller

-0.75

 Airl

-0.73

arij

-0.72

ochond

-0.71

VPN

-0.71

Oracle

-0.70

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

-0.68

POSITIVE LOGITS

 lewd

1.20

 indecent

1.12

 obscene

0.85

 blasp

0.77

 ejac

0.75

Sexual

0.74

 vulgar

0.72

 writ

0.72

 masturb

0.71

uously

0.70

Activations Density 0.016%

words related to inappropriate behavior or language

terms related to indecency and obscenity

No Comments

No Known Activations