INDEX

Explanations

proper nouns or specific identifiers

oai_token-act-pair · gpt-3.5-turbo

phrases related to identification of individuals in reports or news articles

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 plank

-0.68

é¾įå¥ĳå£«

-0.68

 bandwagon

-0.61

 pedest

-0.61

DragonMagazine

-0.59

bench

-0.58

andon

-0.58

gaard

-0.57

 Gleaming

-0.55

ground

-0.55

POSITIVE LOGITS

 anonymity

1.29

 pseudonym

1.11

 initials

0.97

 surn

0.92

 anonymously

0.92

 anonymous

0.90

 surname

0.89

 unnamed

0.88

 name

0.82

 unidentified

0.81

Activations Density 0.122%

proper nouns or specific identifiers

phrases related to identification of individuals in reports or news articles

No Comments

No Known Activations