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

idden

-0.87

tnc

-0.85

proxy

-0.80

rolled

-0.76

role

-0.76

ertodd

-0.75

STATE

-0.74

voice

-0.73

UI

-0.73

uum

-0.72

POSITIVE LOGITS

 charism

0.82

pour

0.72

 rake

0.71

 Pole

0.70

 Rabbi

0.69

 recapt

0.69

Yon

0.68

mort

0.68

Pwr

0.67

 Loft

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.