INDEX

Explanations

the word "un" followed by a verb or adjective

oai_token-act-pair · gpt-3.5-turbo

occurrences of the prefix "Un" or references to "unknown" concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 slit

-0.73

inelli

-0.66

isphere

-0.66

stanbul

-0.66

bsite

-0.64

berman

-0.64

soDeliveryDate

-0.63

dq

-0.63

 rake

-0.62

GI

-0.62

POSITIVE LOGITS

Un

3.21

Un

2.16

un

1.68

un

1.60

Unt

1.59

Und

1.52

Unc

1.49

uns

1.32

UN

1.30

Ung

1.29

Activations Density 0.016%

the word "un" followed by a verb or adjective

occurrences of the prefix "Un" or references to "unknown" concepts

No Comments

No Known Activations