INDEX

Explanations

terms related to news articles and journalism, such as continued, article, story, and photograph

oai_token-act-pair · gpt-3.5-turbo

headings and article indicators within a document

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

iHUD

-0.19

ĺħ

-0.19

é¾įå¥ĳå£«

-0.18

 DPRK

-0.18

 Trotsky

-0.18

 Zoro

-0.18

Footnote

-0.17

Merit

-0.17

IDF

-0.17

©¶æ

-0.17

POSITIVE LOGITS

sheets

0.21

=-=-

0.20

 href

0.19

ookie

0.17

bledon

0.17

izon

0.16

assetsadobe

0.16

quickShipAvailable

0.16

 IMAGES

0.16

jri

0.16

Activations Density 11.798%

terms related to news articles and journalism, such as continued, article, story, and photograph

headings and article indicators within a document

No Comments

No Known Activations