INDEX

Explanations

words related to violent physical actions, particularly stabbings

oai_token-act-pair · gpt-3.5-turbo

references to stabbing incidents and associated injuries

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

oS

-0.73

XM

-0.70

 Cosmos

-0.70

Scot

-0.66

ullivan

-0.65

 Geographic

-0.65

 Seller

-0.64

 Afric

-0.64

VO

-0.64

 Organization

-0.63

POSITIVE LOGITS

 wounds

1.08

lished

1.00

 stabbing

0.87

 slit

0.86

 stab

0.85

 throats

0.83

 rampage

0.82

 stabbed

0.82

 dagger

0.81

 knife

0.80

Activations Density 0.045%

words related to violent physical actions, particularly stabbings

references to stabbing incidents and associated injuries

No Comments

No Known Activations