INDEX

Explanations

mentions of nuclear weapons

oai_token-act-pair · gpt-3.5-turbo

references to nuclear weapons and related programs

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

idas

-0.78

idav

-0.77

soType

-0.73

details

-0.68

 Rough

-0.66

felt

-0.66

 Barber

-0.66

 Parish

-0.66

 Agric

-0.66

eret

-0.66

POSITIVE LOGITS

 plutonium

1.06

utonium

0.97

iferation

0.96

 dismant

0.96

 submarines

0.95

 warheads

0.92

uclear

0.89

 submarine

0.87

 arsen

0.86

 detonated

0.85

Activations Density 0.089%

mentions of nuclear weapons

references to nuclear weapons and related programs

No Comments

No Known Activations