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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

UME

-0.68

 VIDEOS

-0.64

ebus

-0.64

alez

-0.62

ams

-0.61

 pounded

-0.60

iosis

-0.60

 chanted

-0.59

unts

-0.59

 Classics

-0.59

POSITIVE LOGITS

fac

0.70

metic

0.68

ģĸ

0.62

ris

0.61

lift

0.60

terness

0.60

forth

0.60

rieve

0.59

onest

0.59

 coerc

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.