INDEX

Explanations

phrases expressing knowledge or awareness

oai_token-act-pair · gpt-3.5-turbo

instances of the word "knew" and variations of it, indicating familiarity or prior knowledge

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

otion

-0.77

pex

-0.72

phrine

-0.71

area

-0.70

olid

-0.70

 rÃ©

-0.67

ulum

-0.67

 Yugoslavia

-0.64

por

-0.64

 ILCS

-0.63

POSITIVE LOGITS

 beforehand

0.97

ledged

0.83

lessly

0.80

bryce

0.73

nothing

0.72

lege

0.72

saf

0.70

 instinctively

0.69

($)

0.69

footed

0.67

Activations Density 0.032%

phrases expressing knowledge or awareness

instances of the word "knew" and variations of it, indicating familiarity or prior knowledge

No Comments

No Known Activations