INDEX

Explanations

words related to safety and security concerns

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

eric

-0.96

yss

-0.83

issance

-0.82

gres

-0.79

xual

-0.77

nant

-0.76

erers

-0.76

naire

-0.75

erer

-0.75

eday

-0.74

POSITIVE LOGITS

 precautions

1.09

 hazards

1.02

net

0.97

 hazard

0.95

 precaution

0.89

 valve

0.87

 nets

0.87

 valves

0.85

 margins

0.81

 concerns

0.80

Activations Density 0.037%

words related to safety and security concerns

No Comments

No Known Activations