INDEX

Explanations

mentions of political terms such as "term", "ran", "seek" in the context of reelection or multiple terms

oai_token-act-pair · gpt-3.5-turbo

references to political terms and election cycles

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

ICLE

-0.80

rote

-0.77

erie

-0.72

riad

-0.68

ergic

-0.67

yrics

-0.67

ortium

-0.66

rotein

-0.66

LET

-0.65

chy

-0.64

POSITIVE LOGITS

abilities

0.91

ability

0.87

mate

0.84

interstitial

0.81

acity

0.80

term

0.79

cap

0.78

 mate

0.78

marks

0.72

long

0.71

Activations Density 0.016%

mentions of political terms such as "term", "ran", "seek" in the context of reelection or multiple terms

references to political terms and election cycles

No Comments

No Known Activations