INDEX

Explanations

phrases related to violations or breaches of rules, laws, or standards

oai_token-act-pair · gpt-3.5-turbo

instances of the word "violate" and its variations related to legal or ethical breaches

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

iris

-0.82

igers

-0.71

acs

-0.69

rich

-0.68

DragonMagazine

-0.67

pring

-0.67

cephal

-0.66

ppel

-0.66

rians

-0.66

rising

-0.65

POSITIVE LOGITS

 laws

1.11

 confidentiality

1.08

 regulations

1.07

 curfew

1.06

 strict

1.01

 Rule

0.98

 guidelines

0.98

 constitutional

0.97

 treaties

0.97

 prohibitions

0.95

Activations Density 0.115%

phrases related to violations or breaches of rules, laws, or standards

instances of the word "violate" and its variations related to legal or ethical breaches

No Comments

No Known Activations