INDEX

Explanations

phrases related to legal charges

oai_token-act-pair · gpt-3.5-turbo

references to degrees of crime in legal contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

isu

-0.88

soDeliveryDate

-0.85

raltar

-0.81

 Pengu

-0.80

 tremend

-0.80

obyl

-0.76

WOR

-0.75

UNCH

-0.75

sels

-0.73

TAG

-0.73

POSITIVE LOGITS

 manslaughter

0.95

 felony

0.92

degree

0.92

 arson

0.86

rees

0.82

 murder

0.78

 indictment

0.77

 acquaintance

0.76

edly

0.76

 misdemeanor

0.74

Activations Density 0.010%

phrases related to legal charges

references to degrees of crime in legal contexts

No Comments

No Known Activations