INDEX

Explanations

phrases including abbreviations typically found in news or formal writing

oai_token-act-pair · gpt-3.5-turbo

punctuations and sentence endings, indicating the structure of writing

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

itably

-0.95

escription

-0.90

retty

-0.88

 referen

-0.87

acebook

-0.87

irlf

-0.83

ocument

-0.77

mercial

-0.76

ividually

-0.74

efeated

-0.74

POSITIVE LOGITS

NetMessage

1.00

CLASSIFIED

0.95

WAYS

0.88

 largeDownload

0.86

 Lumpur

0.77

natureconservancy

0.77

 toile

0.76

namese

0.75

SELECT

0.73

INESS

0.70

Activations Density 1.980%

phrases including abbreviations typically found in news or formal writing

punctuations and sentence endings, indicating the structure of writing

No Comments

No Known Activations