INDEX

Explanations

references to executives, leadership positions, or corporate roles

oai_token-act-pair · gpt-3.5-turbo

references to executives and executive roles

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

FACE

-0.83

 Brist

-0.71

theless

-0.71

FUL

-0.69

TPS

-0.69

 Franklin

-0.68

 Sense

-0.67

 Telegram

-0.66

 Twain

-0.66

 Weather

-0.66

POSITIVE LOGITS

ution

1.51

utions

1.41

utive

1.37

uting

1.28

utable

1.22

uted

1.14

exec

1.08

utes

1.01

ination

1.01

utor

1.00

Activations Density 0.007%

references to executives, leadership positions, or corporate roles

references to executives and executive roles

No Comments

No Known Activations