INDEX

Explanations

exclamatory words expressing strong emotions, such as surprise or disbelief

oai_token-act-pair · gpt-3.5-turbo

expressions of disbelief or sarcasm

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

•layer 5 - surprise

No Comments

Negative Logits

arthy

-0.75

icipated

-0.71

enary

-0.71

adr

-0.69

athered

-0.64

bett

-0.63

fold

-0.63

idas

-0.62

sylv

-0.61

eded

-0.60

POSITIVE LOGITS

 Seriously

0.84

zers

0.84

 dunno

0.79

 kidding

0.78

FTWARE

0.77

Seriously

0.75

 Helpful

0.75

tho

0.73

?!

0.70

essage

0.69

Activations Density 0.055%

exclamatory words expressing strong emotions, such as surprise or disbelief

expressions of disbelief or sarcasm

No Comments

No Known Activations