INDEX

Explanations

references to hacking techniques or activities

oai_token-act-pair · gpt-3.5-turbo

instances of the word "hack" and related variations indicating hacking activities or themes

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

 concess

-0.70

BuyableInstoreAndOnline

-0.69

âĢ¢âĢ¢âĢ¢âĢ¢

-0.68

 Archdemon

-0.67

reply

-0.67

 Presence

-0.66

 Veter

-0.64

oplan

-0.62

xus

-0.61

ãħĭ

-0.61

POSITIVE LOGITS

intosh

1.39

athon

1.20

hered

1.00

nik

0.94

aday

0.91

lar

0.84

herer

0.83

enberg

0.83

sonian

0.82

busters

0.81

Activations Density 0.016%

references to hacking techniques or activities

instances of the word "hack" and related variations indicating hacking activities or themes

No Comments

No Known Activations