INDEX

Explanations

person or entity identities in relation to possible actions or events

oai_token-act-pair · gpt-3.5-turbo

references to potential suspects or perpetrators involved in criminal activities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ldom

-0.85

VERTISEMENT

-0.82

rules

-0.75

utenberg

-0.73

staking

-0.73

bane

-0.73

Market

-0.72

 winters

-0.72

omics

-0.71

Reviewed

-0.71

POSITIVE LOGITS

 perpetrator

1.62

 perpetrators

1.53

 culprit

1.52

 assailant

1.45

 attacker

1.44

 assailants

1.40

 gunman

1.33

 attackers

1.32

 suspect

1.30

 sender

1.29

Activations Density 0.286%

person or entity identities in relation to possible actions or events

references to potential suspects or perpetrators involved in criminal activities

No Comments

No Known Activations