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

 Dixon

-0.73

 Pitt

-0.66

board

-0.64

OTH

-0.64

 offic

-0.62

 Breast

-0.62

 prior

-0.62

 bunk

-0.62

 Cook

-0.61

 present

-0.61

POSITIVE LOGITS

ãĤ´ãĥ³

1.09

0.86

EStream

0.83

yssey

0.82

maxwell

0.79

catch

0.78

 Parables

0.76

=~

0.75

aliation

0.75

å¹

0.75

Activations Density 0.000%

No Known Activations

This feature has no known activations.