INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

hran

-0.76

lez

-0.73

 sugg

-0.71

orius

-0.71

<?

-0.70

 conclud

-0.67

ppel

-0.66

idon

-0.65

orious

-0.65

 myster

-0.64

POSITIVE LOGITS

day

0.90

IAS

0.73

DAY

0.70

 Indy

0.69

 Azerbaijan

0.66

ç¥ŀ

0.66

 Indianapolis

0.66

IFT

0.65

 Pastebin

0.65

 Scale

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.