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

SPONSORED

-0.89

åĤ

-0.72

ATING

-0.70

ãĤ¼ãĤ¦ãĤ¹

-0.69

ARGET

-0.68

alties

-0.67

obil

-0.65

Assembly

-0.64

ocrine

-0.64

ched

-0.64

POSITIVE LOGITS

uria

0.73

 Antar

0.71

quart

0.70

'/

0.64

berman

0.64

Roy

0.64

 bolted

0.63

ofer

0.63

buquerque

0.59

 Oath

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.