INDEX

Explanations

the name "Julian"

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "Julian"

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

NetMessage

-0.75

liga

-0.73

tml

-0.73

DIV

-0.73

holders

-0.71

stakes

-0.70

close

-0.68

holder

-0.67

doors

-0.67

skirts

-0.67

POSITIVE LOGITS

 Assange

1.19

iets

0.83

ity

0.82

 Julian

0.82

 Fellow

0.80

sson

0.76

athan

0.75

 Caesar

0.72

 Abram

0.71

Tay

0.68

Activations Density 0.009%

the name "Julian"

mentions of the name "Julian"

No Comments

No Known Activations