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

 Pars

-0.77

 Hier

-0.66

 denomin

-0.65

PAC

-0.64

externalActionCode

-0.64

ANG

-0.63

ned

-0.60

 Modified

-0.60

 presiding

-0.59

 Hanson

-0.59

POSITIVE LOGITS

ride

0.88

otine

0.85

udden

0.78

axy

0.76

aturdays

0.76

renheit

0.75

iday

0.74

ogue

0.73

cade

0.72

fters

0.72

Activations Density 0.000%

No Known Activations

This feature has no known activations.