INDEX

Explanations

abbreviations related to a security or intelligence agency, like "MI."

oai_token-act-pair · gpt-3.5-turbo

mentions of the abbreviation "MI" in various contexts

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

ional

-0.88

holders

-0.78

ding

-0.78

\\\\\\\\

-0.76

wagen

-0.75

holder

-0.74

size

-0.74

mans

-0.73

sword

-0.71

language

-0.70

POSITIVE LOGITS

AMI

1.22

PS

1.16

HAEL

1.10

KE

1.03

LE

0.87

OSH

0.86

uppet

0.83

XT

0.82

MI

0.79

LEY

0.78

Activations Density 0.013%

abbreviations related to a security or intelligence agency, like "MI."

mentions of the abbreviation "MI" in various contexts

No Comments

No Known Activations