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

obin

-0.75

estinal

-0.72

iterranean

-0.71

rontal

-0.69

PLIED

-0.67

anamo

-0.65

 Niet

-0.65

gnu

-0.65

CLUS

-0.65

stract

-0.64

POSITIVE LOGITS

as

1.13

As

0.69

 brushes

0.68

As

0.62

eda

0.60

ida

0.56

AS

0.56

as

0.56

akens

0.55

 floors

0.55

Activations Density 0.000%

No Known Activations

This feature has no known activations.