INDEX

Explanations

mentions related to dangerous individuals in news articles

oai_token-act-pair · gpt-3.5-turbo

references to individuals involved in violent incidents, particularly gunmen and shooters

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

arb

-0.80

urb

-0.68

cess

-0.67

Transaction

-0.66

Fut

-0.66

cles

-0.65

 Mining

-0.63

times

-0.62

urn

-0.62

 Highlands

-0.61

POSITIVE LOGITS

 shooter

0.98

 Shooter

0.96

 shooters

0.92

powder

0.80

 gunman

0.78

ÃĽ

0.77

linger

0.75

 Paddock

0.74

 shooting

0.74

nikov

0.73

Activations Density 0.035%

mentions related to dangerous individuals in news articles

references to individuals involved in violent incidents, particularly gunmen and shooters

No Comments

No Known Activations