INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

oret

-0.78

tein

-0.74

essen

-0.70

ieth

-0.70

ively

-0.69

orno

-0.69

rament

-0.68

ua

-0.67

minster

-0.66

ilde

-0.65

POSITIVE LOGITS

beard

0.72

stocks

0.67

byss

0.66

å¤©

0.65

nown

0.63

ppelin

0.63

?:

0.62

Thor

0.62

HUD

0.61

Atl

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.