INDEX

Explanations

words related to personal names, particularly the name "Joel"

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

hib

-0.81

dayName

-0.77

omething

-0.72

FACE

-0.70

forward

-0.66

DragonMagazine

-0.65

BOOK

-0.63

luaj

-0.62

hire

-0.61

 gestation

-0.60

POSITIVE LOGITS

McH

1.14

aquin

0.96

ovic

0.85

owitz

0.84

lein

0.84

 Goodman

0.83

aqu

0.82

oad

0.77

 Rubin

0.77

 Wein

0.76

Activations Density 0.033%

words related to personal names, particularly the name "Joel"

No Comments

No Known Activations