INDEX

Explanations

descriptions or mentions of injuries

oai_token-act-pair · gpt-3.5-turbo

mentions of injuries or harm caused to individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

gency

-0.79

gran

-0.71

perm

-0.70

minist

-0.69

ingen

-0.65

 ellipt

-0.63

 algorithm

-0.62

arten

-0.62

SpaceEngineers

-0.61

ramid

-0.61

POSITIVE LOGITS

jured

0.88

 survivors

0.85

 Survivors

0.81

 injured

0.80

 victims

0.79

 bystanders

0.78

adoes

0.76

 wounded

0.76

 injuring

0.75

../

0.75

Activations Density 0.022%

descriptions or mentions of injuries

mentions of injuries or harm caused to individuals

No Comments

No Known Activations