INDEX

Explanations

references to the concept of 'good'

oai_token-act-pair · gpt-3.5-turbo

references to food

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

UGE

-0.77

 Constantin

-0.74

BuyableInstoreAndOnline

-0.73

Ago

-0.71

MIA

-0.69

USE

-0.68

ASED

-0.66

chal

-0.66

UNCH

-0.66

asper

-0.66

POSITIVE LOGITS

ood

1.21

edly

0.93

ed

0.92

iership

0.91

lers

0.90

rill

0.89

les

0.88

ividual

0.87

ler

0.87

skin

0.86

Activations Density 0.016%

references to the concept of 'good'

references to food

No Comments

No Known Activations