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

Jew

-0.71

 Kinder

-0.69

 Vancouver

-0.68

WA

-0.62

 Federation

-0.61

 wise

-0.61

 Tide

-0.61

 regener

-0.61

pag

-0.60

 Principal

-0.60

POSITIVE LOGITS

ãĤ¼ãĤ¦ãĤ¹

0.91

enza

0.91

onyms

0.85

agues

0.85

uers

0.83

assian

0.80

uminium

0.79

utenberg

0.78

olars

0.77

avorite

0.77

Activations Density 0.000%

No Known Activations

This feature has no known activations.