INDEX

Explanations

phrases containing the word "-wise" which indicate a parameter or manner of doing something

oai_token-act-pair · gpt-3.5-turbo

phrases related to ratings or evaluations of various forms of media and experiences

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

Not in Any Lists

No Comments

Negative Logits

ĸļ

-0.73

bern

-0.70

bley

-0.67

BLE

-0.67

BAT

-0.66

è¦ļéĨĴ

-0.66

onde

-0.66

rose

-0.65

nikov

-0.64

 Noir

-0.63

POSITIVE LOGITS

lihood

1.07

wise

1.05

ably

0.82

 guiIcon

0.78

 fixme

0.76

 è£ıè

0.75

ly

0.73

é¾įå¥ĳå£«

0.72

liness

0.71

ever

0.68

Activations Density 0.006%

phrases containing the word "-wise" which indicate a parameter or manner of doing something

phrases related to ratings or evaluations of various forms of media and experiences

No Comments

No Known Activations