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

agar

-0.72

agy

-0.71

erness

-0.68

auga

-0.64

oder

-0.63

 impede

-0.63

 encount

-0.63

iour

-0.62

adin

-0.62

ospace

-0.61

POSITIVE LOGITS

 Cosponsors

0.75

blogspot

0.73

 Photos

0.72

↵

0.68

 Rasmussen

0.65

î

0.64

↵Âł

0.64

 Michele

0.62

Pinterest

0.61

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.