INDEX

Explanations

phrases related to familiar topics or concepts, especially in the context of equivalences and comparisons

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate harm or danger to individuals

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

tarians

-0.40

cigarettes

-0.40

estern

-0.36

DragonMagazine

-0.36

unctions

-0.36

 Canaver

-0.35

Â¶

-0.34

itars

-0.34

cedented

-0.34

 Econom

-0.33

POSITIVE LOGITS

 overhead

0.34

 interchange

0.33

 nearby

0.32

.""

0.31

 imitation

0.31

 respectively

0.31

}.

0.30

 glor

0.30

 unspecified

0.30

 electro

0.30

Activations Density 7.027%

phrases related to familiar topics or concepts, especially in the context of equivalences and comparisons

phrases that indicate harm or danger to individuals

No Comments

No Known Activations