INDEX

Explanations

terms related to political partisanship

oai_token-act-pair · gpt-3.5-turbo

terms associated with partisanship and political division

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

uras

-0.89

uran

-0.81

ternally

-0.79

enium

-0.77

eon

-0.76

hell

-0.76

ofi

-0.75

ulet

-0.75

aryn

-0.74

oppers

-0.74

POSITIVE LOGITS

 partisans

1.15

 partisan

1.15

 affiliation

0.83

psy

0.82

 aggregation

0.80

 persuasion

0.80

artisan

0.77

 affili

0.77

 elector

0.75

 strugg

0.75

Activations Density 0.007%

terms related to political partisanship

terms associated with partisanship and political division

No Comments

No Known Activations