INDEX

Explanations

references to locations and violent events involving death

oai_token-act-pair · gpt-3.5-turbo

references to locations and situations involving death or violence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

natureconservancy

-0.74

 partName

-0.73

antz

-0.66

itized

-0.66

IRED

-0.66

imil

-0.65

 plus

-0.65

SHIP

-0.63

strap

-0.62

HD

-0.62

POSITIVE LOGITS

 womb

1.01

 woods

1.00

 streets

0.98

 custody

0.97

 vicinity

0.94

 bathroom

0.91

 midst

0.89

 restroom

0.88

 bushes

0.88

 thro

0.86

Activations Density 0.207%

references to locations and violent events involving death

references to locations and situations involving death or violence

No Comments

No Known Activations