INDEX

Explanations

multiple mentions of the name "Roy Moore"

oai_token-act-pair · gpt-3.5-turbo

references to political candidates, particularly Roy Moore and Todd Akin

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

OTS

-0.80

OTAL

-0.78

IPS

-0.69

eric

-0.66

retched

-0.65

ä¸ī

-0.65

 BahÃ¡

-0.64

NIC

-0.62

Liquid

-0.62

 Feder

-0.62

POSITIVE LOGITS

 accus

1.04

 accuser

0.87

milo

0.80

 voters

0.79

 supporters

0.77

 candidacy

0.76

 Voters

0.75

 reelection

0.74

 strongh

0.72

 stances

0.72

Activations Density 0.042%

multiple mentions of the name "Roy Moore"

references to political candidates, particularly Roy Moore and Todd Akin

No Comments

No Known Activations