INDEX

Explanations

mentions of winning in political contexts, especially related to elections

oai_token-act-pair · gpt-3.5-turbo

references to winning elections and victories 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 @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

DragonMagazine

-0.77

dash

-0.69

assetsadobe

-0.67

 VIDEOS

-0.65

 Minotaur

-0.65

 pros

-0.64

obyl

-0.64

Iterator

-0.64

bek

-0.63

atari

-0.62

POSITIVE LOGITS

 reelection

1.32

 majorities

1.07

 landsl

1.00

 landslide

0.96

 primaries

0.95

 convinc

0.94

0.92

 statewide

0.92

 decisively

0.91

 elections

0.90

Activations Density 0.085%

mentions of winning in political contexts, especially related to elections

references to winning elections and victories in a political context

No Comments

No Known Activations