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

 Huffington

-0.64

 Bard

-0.63

 Buddha

-0.63

 Subject

-0.61

 Franklin

-0.60

èĢħ

-0.60

 Strauss

-0.58

 STAT

-0.57

hematic

-0.56

 subject

-0.55

POSITIVE LOGITS

glas

0.74

IX

0.73

arantine

0.72

IT

0.71

rench

0.71

tu

0.70

saf

0.70

it

0.69

Ukraine

0.68

DragonMagazine

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.