INDEX

Explanations

mentions of "Short" as part of a proper noun, possibly referring to names like titles, names, or organizations

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Short" used in various contexts

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

 allegiance

-0.69

avia

-0.68

roph

-0.66

 attest

-0.64

lia

-0.62

vet

-0.61

 insur

-0.60

 observ

-0.60

agara

-0.59

ront

-0.59

POSITIVE LOGITS

 Short

3.71

Short

2.94

short

2.55

 short

1.95

 Long

1.52

 shortest

1.40

Long

1.37

 shorter

1.28

 shortcuts

1.26

 shorten

1.23

Activations Density 0.024%

mentions of "Short" as part of a proper noun, possibly referring to names like titles, names, or organizations

instances of the word "Short" used in various contexts

No Comments

No Known Activations