INDEX

Explanations

common introductory phrases in news articles

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the presence of sections or paragraphs in a document

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 unemploy

-0.79

reon

-0.72

 combatants

-0.70

$$

-0.66

 lett

-0.65

 nurt

-0.63

 actionGroup

-0.63

 looms

-0.62

channelAvailability

-0.62

MpServer

-0.62

POSITIVE LOGITS

 addition

1.28

cluding

1.24

spect

1.21

ventory

1.17

 contrast

1.15

 spite

1.12

vasive

1.11

mates

1.10

cluded

1.10

jured

1.09

Activations Density 0.123%

common introductory phrases in news articles

phrases indicating the presence of sections or paragraphs in a document

No Comments

No Known Activations