INDEX

Explanations

the word "or" with a strong activation value

oai_token-act-pair · gpt-3.5-turbo

phrases indicating alternatives or choices

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

 Pony

-0.90

ocracy

-0.76

ocrats

-0.74

ocrat

-0.72

achine

-0.71

ulhu

-0.70

ascus

-0.70

ARDS

-0.68

 Rocket

-0.66

efer

-0.66

POSITIVE LOGITS

chard

1.35

Else

1.35

nam

1.29

acles

1.27

ifice

1.23

acle

1.23

nery

1.21

chid

1.14

 otherwise

1.07

 else

1.05

Activations Density 0.145%

the word "or" with a strong activation value

phrases indicating alternatives or choices

No Comments

No Known Activations