INDEX

Explanations

mentions of political office holders

oai_token-act-pair · gpt-3.5-turbo

references to incumbents in a political context

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

wagen

-0.95

 Flavoring

-0.83

gas

-0.83

ooter

-0.82

oho

-0.82

kie

-0.82

orgetown

-0.79

stocks

-0.77

hod

-0.76

okin

-0.76

POSITIVE LOGITS

 incumbent

1.07

 challenger

0.89

 incumb

0.88

 frontrunner

0.82

 loser

0.79

 challengers

0.77

 stal

0.73

foe

0.73

 tenant

0.73

 strongh

0.72

Activations Density 0.023%

mentions of political office holders

references to incumbents in a political context

No Comments

No Known Activations