INDEX

Explanations

mentions of and references to political affiliations, particularly the Democratic Party

oai_token-act-pair · gpt-3.5-turbo

references to the Democratic Party and its members

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

weather

-0.76

obyl

-0.76

illin

-0.75

iddles

-0.72

laus

-0.71

aning

-0.71

urious

-0.70

itta

-0.70

iffs

-0.69

iries

-0.69

POSITIVE LOGITS

ization

0.74

naissance

0.72

izable

0.71

 Party

0.71

 frontrunner

0.70

hood

0.69

 stronghold

0.69

 politician

0.67

 standby

0.65

 bund

0.65

Activations Density 0.016%

mentions of and references to political affiliations, particularly the Democratic Party

references to the Democratic Party and its members

No Comments

No Known Activations