INDEX

Explanations

years mentioned in historical contexts

oai_token-act-pair · gpt-3.5-turbo

instances of the word "In" followed by a numerical reference or date

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 lett

-0.70

 spoil

-0.70

 behav

-0.69

 ingred

-0.67

 looms

-0.63

 lodged

-0.63

 combatants

-0.63

 selves

-0.62

 babe

-0.62

 outwe

-0.61

POSITIVE LOGITS

ventory

1.38

cluding

1.37

clusion

1.25

juries

1.23

 addition

1.21

jection

1.20

jured

1.18

jury

1.16

cluded

1.15

 lieu

1.15

Activations Density 0.147%

years mentioned in historical contexts

instances of the word "In" followed by a numerical reference or date

No Comments

No Known Activations