INDEX

Explanations

mentions of a specific section or type of written content, possibly related to opinions or articles

oai_token-act-pair · gpt-3.5-turbo

mentions of columns or articles related to opinions

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

CRIP

-0.76

 Romanian

-0.71

à¹

-0.71

qual

-0.71

lder

-0.70

gets

-0.63

ãĤ©

-0.62

zie

-0.62

byss

-0.61

EGA

-0.61

POSITIVE LOGITS

 columns

0.98

 headers

0.88

 inches

0.86

ists

0.85

 column

0.83

antry

0.83

ftime

0.79

iers

0.78

column

0.78

ials

0.76

Activations Density 0.016%

mentions of a specific section or type of written content, possibly related to opinions or articles

mentions of columns or articles related to opinions

No Comments

No Known Activations