INDEX

Explanations

mentions of political figures or topics

oai_token-act-pair · gpt-3.5-turbo

references to political topics

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

 Warrant

-0.74

 warrants

-0.68

 pity

-0.67

 Carbuncle

-0.66

 recall

-0.62

 Gutenberg

-0.62

PORT

-0.61

MQ

-0.61

Ved

-0.61

FACE

-0.60

POSITIVE LOGITS

icians

1.64

ician

1.50

ifact

1.26

ically

1.22

eness

1.17

icial

1.09

ileaks

0.99

ique

0.97

icking

0.95

kov

0.93

Activations Density 0.026%

mentions of political figures or topics

references to political topics

No Comments

No Known Activations