INDEX

Explanations

locations or institutions with a specific word in them

oai_token-act-pair · gpt-3.5-turbo

locations mentioned in an academic context

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

DragonMagazine

-0.77

anos

-0.68

sed

-0.63

 taco

-0.62

 tacos

-0.61

eto

-0.61

 Kislyak

-0.61

psons

-0.60

 Mari

-0.60

anches

-0.59

POSITIVE LOGITS

 England

1.51

England

1.48

UK

1.39

 Australia

1.28

 Ireland

1.23

Australia

1.20

 Ontario

1.20

 Sweden

1.19

 Scotland

1.19

 Denmark

1.18

Activations Density 0.104%

locations or institutions with a specific word in them

locations mentioned in an academic context

No Comments

No Known Activations