INDEX

Explanations

mentions of close relationships or connections between people or entities

oai_token-act-pair · gpt-3.5-turbo

references to close relationships or connections

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

ICAN

-0.84

AIN

-0.69

ULE

-0.69

mA

-0.68

NZ

-0.68

icist

-0.67

Ø©

-0.66

ican

-0.66

oeuv

-0.65

alian

-0.64

POSITIVE LOGITS

 minded

0.96

 proximity

0.90

 confinement

0.86

sighted

0.79

 confid

0.78

 relations

0.77

 doors

0.76

 shave

0.76

 ties

0.76

minded

0.73

Activations Density 0.025%

mentions of close relationships or connections between people or entities

references to close relationships or connections

No Comments

No Known Activations