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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Reply

-0.81

 seiz

-0.73

Daily

-0.72

Rated

-0.69

 rundown

-0.67

 cuff

-0.65

ilts

-0.61

Frames

-0.61

Release

-0.61

 heel

-0.61

POSITIVE LOGITS

Ave

0.76

 Pike

0.72

 Barnett

0.71

lake

0.70

Ari

0.70

 Patterson

0.68

igon

0.68

cil

0.66

 Lans

0.66

 Peterson

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.