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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Reviewed

-0.83

APH

-0.77

Louis

-0.75

sic

-0.75

CI

-0.74

ihar

-0.73

Virgin

-0.72

 contrace

-0.67

utical

-0.67

classified

-0.66

POSITIVE LOGITS

 Yamato

0.70

syn

0.69

éĹĺ

0.68

oresc

0.67

 Bers

0.67

awks

0.65

 impe

0.63

dale

0.62

 Clock

0.62

 Anime

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.