INDEX

Explanations

references to fictional or mythical artifacts or objects

oai_token-act-pair · gpt-3.5-turbo

phrases that signify ownership or association with certain items or concepts

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

BILL

-0.85

 weekday

-0.74

 commercially

-0.73

 drawer

-0.70

 plug

-0.69

 pled

-0.69

fully

-0.68

bom

-0.66

picture

-0.66

 nodd

-0.66

POSITIVE LOGITS

 Eternity

1.20

 Ages

1.16

 Destruction

1.15

 Sacrifice

1.14

 Souls

1.13

 Darkness

1.10

 Tears

1.10

 Corruption

1.09

 Wisdom

1.09

 Dreams

1.08

Activations Density 0.097%

references to fictional or mythical artifacts or objects

phrases that signify ownership or association with certain items or concepts

No Comments

No Known Activations