INDEX

Explanations

the word "sure" at a high activation level

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "I'm sure"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ufact

-0.84

LAN

-0.73

inese

-0.73

inventoryQuantity

-0.69

INT

-0.66

bank

-0.65

DragonMagazine

-0.64

RIP

-0.64

 Flavoring

-0.62

UCH

-0.61

POSITIVE LOGITS

 enough

0.72

terday

0.71

 sure

0.71

 Virtue

0.71

ties

0.69

comed

0.66

ures

0.63

heon

0.62

 tomat

0.62

 Gamble

0.62

Activations Density 0.026%

the word "sure" at a high activation level

instances of the phrase "I'm sure"

No Comments

No Known Activations