INDEX

Explanations

words related to legal or illegal actions, specifically focusing on the concept of "lawful" versus "unlawful"

oai_token-act-pair · gpt-3.5-turbo

terms related to legal legitimacy and compliance

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

 Tycoon

-0.85

PsyNetMessage

-0.77

Rin

-0.72

ogy

-0.71

Mini

-0.71

utm

-0.69

ulative

-0.68

 Solitaire

-0.68

lasses

-0.67

WAR

-0.67

POSITIVE LOGITS

 lawful

0.90

 unlawful

0.89

 unlawfully

0.88

theless

0.82

 Liberties

0.81

 lawfully

0.81

estial

0.79

00200000

0.78

urnal

0.77

cy

0.73

Activations Density 0.011%

words related to legal or illegal actions, specifically focusing on the concept of "lawful" versus "unlawful"

terms related to legal legitimacy and compliance

No Comments

No Known Activations