INDEX

Explanations

mentions of gifts or presents

oai_token-act-pair · gpt-3.5-turbo

references to gifts and presents

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

tics

-0.83

Dick

-0.71

idents

-0.69

 Occupations

-0.68

 constitu

-0.64

 Ibid

-0.63

arios

-0.62

 Modified

-0.62

å°Ĩ

-0.62

sei

-0.59

POSITIVE LOGITS

 gift

1.06

 baskets

1.05

 gifts

1.01

giving

0.98

 basket

0.91

apult

0.89

wra

0.87

 Gift

0.86

bag

0.86

 Gifts

0.86

Activations Density 0.019%

mentions of gifts or presents

references to gifts and presents

No Comments

No Known Activations