INDEX

Explanations

phrases prompting the viewer to watch something

oai_token-act-pair · gpt-3.5-turbo

phrases that encourage watching videos or content

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

 Germ

-0.73

room

-0.64

ib

-0.62

Ram

-0.62

Ess

-0.62

 Alban

-0.61

 param

-0.60

Lin

-0.60

 Camb

-0.60

ef

-0.59

POSITIVE LOGITS

 WATCH

4.10

WATCH

1.47

 VIDE

1.32

 VIDEOS

1.24

 VIDEO

1.22

 WATCHED

1.22

SEE

1.16

 READ

1.15

 IMAGES

1.11

 Watch

1.11

Activations Density 0.013%

phrases prompting the viewer to watch something

phrases that encourage watching videos or content

No Comments

No Known Activations