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

lyr

-0.78

lain

-0.73

ierrez

-0.72

olicy

-0.71

uese

-0.64

aer

-0.63

emort

-0.63

ebin

-0.62

zech

-0.62

 appetite

-0.61

POSITIVE LOGITS

groups

0.72

selves

0.72

ÙĦ

0.71

boxes

0.70

Austral

0.68

pieces

0.67

 Houses

0.67

 Committees

0.66

oad

0.66

umes

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.