INDEX

Explanations

mentions of the word "Mail" in the text

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Mail"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

razil

-0.75

issan

-0.70

osed

-0.69

iscopal

-0.64

PsyNetMessage

-0.63

cffffcc

-0.63

structed

-0.63

displayText

-0.62

sem

-0.62

itial

-0.61

POSITIVE LOGITS

bag

1.22

bags

1.13

 Mail

0.95

mail

0.92

boxes

0.92

Mail

0.89

box

0.88

ing

0.86

 Online

0.82

Box

0.75

Activations Density 0.008%

mentions of the word "Mail" in the text

instances of the word "Mail"

No Comments

No Known Activations