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

oton

-0.78

aunder

-0.77

atis

-0.76

thumbnails

-0.75

ijah

-0.73

adoes

-0.70

zos

-0.70

pite

-0.69

undown

-0.67

awatts

-0.65

POSITIVE LOGITS

////

0.65

FN

0.64

Ã»

0.64

 Aren

0.62

 Sorceress

0.60

 Sakuya

0.60

 Barbie

0.59

rist

0.58

pires

0.58

lav

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.