INDEX

Explanations

Twitter usernames preceded by the "@" symbol

oai_token-act-pair · gpt-3.5-turbo

mentions of social media handles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

oslov

-0.70

arette

-0.65

 cigarette

-0.61

 gambling

-0.60

 cigarettes

-0.59

 nicotine

-0.58

Sov

-0.58

Ö¼

-0.57

 eaves

-0.57

Footnote

-0.55

POSITIVE LOGITS

3.73

2.32

"@

2.27

(@

1.91

1.62

 twitter

1.43

 tweets

1.32

 retweet

1.31

 tweeting

1.28

 Twitter

1.27

Activations Density 0.010%

Twitter usernames preceded by the "@" symbol

mentions of social media handles

No Comments

No Known Activations