INDEX

Explanations

terms related to the word "Hack"

oai_token-act-pair · gpt-3.5-turbo

references to hacking or related activities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 occas

-0.77

 compr

-0.75

iatrics

-0.72

rul

-0.70

 concess

-0.69

reply

-0.67

ãħĭ

-0.67

accompan

-0.64

 Demonic

-0.64

 polygamy

-0.63

POSITIVE LOGITS

intosh

1.43

athon

1.18

aday

1.08

enberg

1.03

Hack

1.00

busters

0.88

driver

0.86

lings

0.85

ett

0.82

RF

0.82

Activations Density 0.011%

terms related to the word "Hack"

references to hacking or related activities

No Comments

No Known Activations