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

Negative Logits

aned

-0.84

genre

-0.73

alach

-0.67

FY

-0.65

united

-0.63

»Ĵ

-0.63

Done

-0.62

aney

-0.62

 Flavoring

-0.62

 subscribed

-0.61

POSITIVE LOGITS

gain

0.71

 correction

0.71

ilitation

0.64

pound

0.63

sight

0.63

 corrections

0.62

uracy

0.62

llan

0.60

rade

0.60

)|

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.