INDEX

Explanations

phrases related to sensitive topics or information

oai_token-act-pair · gpt-3.5-turbo

references to sensitive topics or issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Wolver

-0.76

 Helsinki

-0.74

AUT

-0.73

ALK

-0.71

mere

-0.68

RON

-0.65

INST

-0.65

YC

-0.65

 Fall

-0.65

AZ

-0.65

POSITIVE LOGITS

 sensitive

1.55

sensitive

1.12

ivities

1.02

 sensit

0.99

ensitive

0.98

 sensitivity

0.95

 proble

0.90

 insensitive

0.85

mble

0.84

 vulner

0.82

Activations Density 0.010%

phrases related to sensitive topics or information

references to sensitive topics or issues

No Comments

No Known Activations