INDEX

Explanations

phrases related to notifying, informing, and alerting others

oai_token-act-pair · gpt-3.5-turbo

references to individuals or entities being informed or notified

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

agne

-0.59

Words

-0.58

xtap

-0.54

ulia

-0.52

evil

-0.52

aband

-0.52

BuyableInstoreAndOnline

-0.51

ternity

-0.51

uner

-0.50

Ô

-0.50

POSITIVE LOGITS

 thereof

1.11

of

1.06

 that

1.02

 about

0.96

that

0.89

of

0.88

OF

0.86

 beforehand

0.81

 otherwise

0.81

about

0.80

Activations Density 0.173%

phrases related to notifying, informing, and alerting others

references to individuals or entities being informed or notified

No Comments

No Known Activations