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

 Nether

-0.63

 Payton

-0.62

 endowed

-0.62

 Myster

-0.61

esan

-0.61

 mund

-0.61

 Akin

-0.60

 Alchemy

-0.59

 Clayton

-0.59

åŃ

-0.58

POSITIVE LOGITS

heads

0.93

agen

0.77

agar

0.70

yss

0.69

Stars

0.65

oÄŁ

0.64

ached

0.63

notations

0.63

eters

0.63

brance

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.