INDEX

Explanations

mentions of political party affiliation, specifically the word 'Democrat'

oai_token-act-pair · gpt-3.5-turbo

mentions of the Democratic Party or references to individuals identified as Democrats

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

laus

-0.90

ounter

-0.78

mberg

-0.73

urious

-0.69

armac

-0.68

atted

-0.67

aning

-0.67

Ku

-0.67

reek

-0.67

utils

-0.65

POSITIVE LOGITS

 Party

0.74

 strongh

0.73

 nominee

0.73

 frontrunner

0.73

 republic

0.72

 bloc

0.72

 bombshell

0.72

naissance

0.71

 stronghold

0.71

ization

0.68

Activations Density 0.008%

mentions of political party affiliation, specifically the word 'Democrat'

mentions of the Democratic Party or references to individuals identified as Democrats

No Comments

No Known Activations