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

 è£ıç

-0.77

 Citiz

-0.70

ismo

-0.68

ISM

-0.67

 Dwar

-0.66

izophren

-0.65

 Pirate

-0.64

rah

-0.63

 Demons

-0.63

 Evangel

-0.62

POSITIVE LOGITS

mark

0.78

 clipboard

0.75

bolt

0.75

wich

0.73

gas

0.72

ascript

0.71

gage

0.69

 licensing

0.68

iferation

0.68

gate

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.