INDEX

Explanations

phrases related to political regimes

oai_token-act-pair · gpt-3.5-turbo

references to authoritarian or oppressive governments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Redd

-0.80

Kent

-0.75

Kin

-0.72

Ort

-0.72

Constructed

-0.72

 Lent

-0.72

Scot

-0.71

 Edinburgh

-0.71

ritch

-0.70

Hir

-0.69

POSITIVE LOGITS

 regime

1.36

 regimes

1.30

 dictator

0.95

 dictatorship

0.92

rul

0.92

aution

0.88

 administ

0.87

ollah

0.87

 governing

0.86

achev

0.86

Activations Density 0.007%

phrases related to political regimes

references to authoritarian or oppressive governments

No Comments

No Known Activations