INDEX

Explanations

Twitter handles and usernames starting with '@'

oai_token-act-pair · gpt-3.5-turbo

mentions of social media interactions and specific individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

azel

-0.93

 mosquito

-0.90

 meter

-0.88

 mosquitoes

-0.85

 leptin

-0.84

fet

-0.83

mun

-0.82

rez

-0.82

ael

-0.81

MTA

-0.81

POSITIVE LOGITS

 Steve

1.98

Steve

1.91

Steven

1.85

 Steven

1.81

 Stevens

1.50

 Stephen

1.49

Stephen

1.43

 Stephenson

1.28

 Stev

1.18

 Gibson

1.10

Activations Density 0.198%

Twitter handles and usernames starting with '@'

mentions of social media interactions and specific individuals

No Comments

No Known Activations