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

aughs

-0.71

é¾įå¥ĳå£«

-0.70

 Dawkins

-0.68

 Aliens

-0.66

 Kinnikuman

-0.66

advertisement

-0.64

 rumors

-0.63

advertising

-0.61

arms

-0.61

 nephew

-0.61

POSITIVE LOGITS

unction

0.91

 Flavoring

0.76

ftime

0.76

Title

0.73

htar

0.70

Task

0.70

ilee

0.70

lly

0.69

phasis

0.68

Spot

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.