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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Seller

-0.75

claimer

-0.72

 palp

-0.68

 Vaughn

-0.68

 Stevens

-0.68

owned

-0.67

ãĤ©

-0.65

 Citation

-0.64

encers

-0.62

iliated

-0.61

POSITIVE LOGITS

 messenger

0.70

eco

0.69

ocalyptic

0.67

gypt

0.66

 billboards

0.64

 poll

0.64

azi

0.63

 festivities

0.62

 transit

0.62

 commentator

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.