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

roman

-0.79

 traged

-0.75

Ĥİ

-0.72

 subscrib

-0.70

owship

-0.70

osure

-0.69

itsch

-0.68

 satell

-0.67

THER

-0.66

ãĤ®

-0.66

POSITIVE LOGITS

_>

0.81

 Ghana

0.71

ãĥ¼ãĥĨ

0.71

kov

0.69

 Lerner

0.65

marg

0.64

 Youth

0.63

 Hispanics

0.62

 Tanzania

0.61

 Uganda

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.