INDEX

Explanations

specific references to the term "Scrolls"

oai_token-act-pair · gpt-3.5-turbo

references to specific games and scrolls within them

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

CVE

-0.72

ocard

-0.69

dit

-0.65

ppers

-0.63

MIL

-0.62

iband

-0.61

 Hyde

-0.60

isher

-0.60

yip

-0.60

Agu

-0.59

POSITIVE LOGITS

 Scrolls

1.41

 scrolls

1.11

 Scroll

0.89

Scroll

0.81

 Mysteries

0.80

 potions

0.78

 Wonders

0.76

 manuscripts

0.75

uality

0.75

 Rings

0.75

Activations Density 0.012%

specific references to the term "Scrolls"

references to specific games and scrolls within them

No Comments

No Known Activations