INDEX

Explanations

mentions of weapons or items related to violence, such as arms and guns

oai_token-act-pair · gpt-3.5-turbo

references to weapons or armaments

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

 Woodward

-0.81

ãĤ©

-0.80

DAY

-0.74

é¾įå

-0.73

 Sound

-0.70

 Dough

-0.70

IENCE

-0.68

TION

-0.68

 Living

-0.64

KEY

-0.63

POSITIVE LOGITS

ament

1.11

 arms

1.06

arm

0.96

aments

0.94

chair

0.86

guards

0.85

chairs

0.81

ageddon

0.81

 cannons

0.78

 cannon

0.78

Activations Density 0.010%

mentions of weapons or items related to violence, such as arms and guns

references to weapons or armaments

No Comments

No Known Activations