INDEX

Explanations

mentions of specific individuals or groups of people

oai_token-act-pair · gpt-3.5-turbo

references to people involved in crime or incidents

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

raltar

-0.66

aeda

-0.64

ä¸Ĭ

-0.64

actionDate

-0.64

Future

-0.61

united

-0.61

equality

-0.60

Lens

-0.59

 moss

-0.59

 blight

-0.59

POSITIVE LOGITS

 testified

0.88

 complained

0.84

 fled

0.83

's

0.83

 pleaded

0.83

 withdrew

0.81

 thanked

0.80

 apologized

0.80

 reportedly

0.79

hari

0.79

Activations Density 0.223%

mentions of specific individuals or groups of people

references to people involved in crime or incidents

No Comments

No Known Activations