INDEX

Explanations

contractions of the word "are" in sentences

oai_token-act-pair · gpt-3.5-turbo

the contraction "we're"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 concealed

-0.58

att

-0.56

 prior

-0.55

 sund

-0.55

 observation

-0.53

 entry

-0.53

 online

-0.52

 herald

-0.52

 following

-0.52

 abroad

-0.52

POSITIVE LOGITS

're

2.97

've

1.92

'll

1.68

'd

1.46

 aren

1.43

'm

1.41

 weren

1.38

are

1.25

Are

1.22

ARE

1.15

Activations Density 0.034%

contractions of the word "are" in sentences

the contraction "we're"

No Comments

No Known Activations