INDEX

Explanations

mentions of usernames or handles on social media platforms

oai_token-act-pair · gpt-3.5-turbo

references to social media interactions and events

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 pione

-0.88

 earthqu

-0.88

Þ

-0.87

ò

-0.86

ñ

-0.85

ß

-0.82

oreAnd

-0.82

ą

-0.81

Ĝ

-0.80

ActionCode

-0.80

POSITIVE LOGITS

])

0.65

]);

0.65

eme

0.62

visors

0.62

lege

0.59

inar

0.58

romeda

0.58

earable

0.58

wordpress

0.57

office

0.57

Activations Density 0.216%

mentions of usernames or handles on social media platforms

references to social media interactions and events

No Comments

No Known Activations