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

adobe

-0.70

Synopsis

-0.68

Attributes

-0.65

 rede

-0.64

Filename

-0.63

faces

-0.62

ights

-0.61

yout

-0.59

 apes

-0.59

 heels

-0.58

POSITIVE LOGITS

ĸļ

0.83

 syrup

0.70

uten

0.68

andestine

0.66

oustic

0.64

romptu

0.63

anche

0.63

rozen

0.63

staking

0.62

 Cunningham

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.