INDEX

Explanations

Twitter usernames

oai_token-act-pair · gpt-3.5-turbo

social media handles and links

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 textbook

-0.75

 textbooks

-0.68

 overtime

-0.67

 Exhibit

-0.66

 diagrams

-0.63

 manual

-0.61

 Dracula

-0.60

CGI

-0.59

 sights

-0.59

 literature

-0.58

POSITIVE LOGITS

1.51

1.19

twitter

1.12

 twitter

1.07

(@

0.99

twitch

0.98

 Follow

0.97

"@

0.94

Follow

0.91

www

0.90

Activations Density 0.054%

Twitter usernames

social media handles and links

No Comments

No Known Activations