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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

UAE

-0.72

 Niger

-0.72

 appreciated

-0.69

 Maurit

-0.69

 Lauder

-0.67

 Alger

-0.65

 Argent

-0.64

iers

-0.64

 Belg

-0.63

 Algeria

-0.63

POSITIVE LOGITS

rolet

0.82

COMPLE

0.76

axter

0.75

netic

0.72

reprene

0.69

irtual

0.69

zees

0.69

warts

0.68

achine

0.68

anchester

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.