INDEX

Explanations

words that may be related to geographic locations

oai_token-act-pair · gpt-3.5-turbo

verbal forms or modifications typically ending in 'ak', 'ip', 'ie', and various suffix patterns

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

atform

-0.65

ãĤ©

-0.60

 pregn

-0.59

Self

-0.58

DragonMagazine

-0.58

ourced

-0.57

assetsadobe

-0.55

TextColor

-0.55

POL

-0.55

ALLY

-0.54

POSITIVE LOGITS

lein

0.83

onen

0.77

hent

0.76

hurst

0.75

etus

0.75

hower

0.74

haar

0.73

enges

0.73

orius

0.71

gaard

0.70

Activations Density 0.118%

words that may be related to geographic locations

verbal forms or modifications typically ending in 'ak', 'ip', 'ie', and various suffix patterns

No Comments

No Known Activations