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

 Eliot

-0.74

grim

-0.66

dan

-0.65

rub

-0.65

 Coat

-0.65

fax

-0.65

nikov

-0.63

 Palest

-0.63

haz

-0.63

Berry

-0.62

POSITIVE LOGITS

interstitial

0.93

Frames

0.78

erences

0.77

going

0.72

stellar

0.71

ween

0.70

successful

0.69

Downloadha

0.69

initely

0.68

oct

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.