INDEX

Explanations

online profiles with information about users

oai_token-act-pair · gpt-3.5-turbo

references to user profiles and their activity on platforms

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

andro

-0.83

onga

-0.78

oshenko

-0.76

agos

-0.72

banks

-0.71

 GOODMAN

-0.67

flies

-0.65

ouf

-0.64

irlf

-0.64

 Sturgeon

-0.64

POSITIVE LOGITS

 Joined

1.28

 Blog

0.74

 Prediction

0.64

CLASSIFIED

0.64

AZ

0.64

 Rankings

0.64

Aug

0.63

Posts

0.62

Activity

0.62

 Colo

0.62

Activations Density 0.012%

online profiles with information about users

references to user profiles and their activity on platforms

No Comments

No Known Activations