INDEX

Explanations

phrases related to social media and online behavior

oai_token-act-pair · gpt-3.5-turbo

disparaging remarks or offensive comments made online

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

cells

-0.78

asio

-0.76

 Ports

-0.73

EVs

-0.72

ongevity

-0.72

 regeneration

-0.70

 Mechan

-0.70

urion

-0.69

 endurance

-0.69

 Endurance

-0.69

POSITIVE LOGITS

 derogatory

1.47

 slurs

1.35

 insulting

1.34

 lewd

1.28

 objectionable

1.28

 vulgar

1.27

 disrespectful

1.26

 blasp

1.25

 inappropriate

1.24

 insults

1.21

Activations Density 0.898%

phrases related to social media and online behavior

disparaging remarks or offensive comments made online

No Comments

No Known Activations