INDEX

Explanations

food-related terms, specifically "stuffed."

oai_token-act-pair · gpt-3.5-turbo

references to objects that are filled or packed with something

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

externalActionCode

-0.75

APH

-0.73

INAL

-0.70

metic

-0.66

assian

-0.63

casts

-0.63

reviewed

-0.61

ndum

-0.60

ignty

-0.60

 diligence

-0.60

POSITIVE LOGITS

 stuffed

1.03

nesday

0.90

bag

0.89

ulously

0.88

 stuffing

0.87

erton

0.84

adelphia

0.81

bags

0.81

glers

0.80

UFF

0.79

Activations Density 0.021%

food-related terms, specifically "stuffed."

references to objects that are filled or packed with something

No Comments

No Known Activations