INDEX

Explanations

references to hijackings

oai_token-act-pair · gpt-3.5-turbo

references to hijacking events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Solitaire

-0.75

-0.74

 Angels

-0.70

 Staples

-0.69

ç¥ŀ

-0.68

 Pompe

-0.66

 Rite

-0.64

 Izan

-0.63

 Piercing

-0.63

 Fairfax

-0.62

POSITIVE LOGITS

ackers

1.17

acker

1.11

acking

1.02

inx

0.99

hij

0.96

adish

0.92

abis

0.85

intosh

0.81

agic

0.79

acked

0.78

Activations Density 0.022%

references to hijackings

references to hijacking events

No Comments

No Known Activations