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

Negative Logits

 Avenger

-0.64

Magn

-0.64

paren

-0.64

mun

-0.61

itÃ©

-0.60

 Substance

-0.60

 ILCS

-0.59

jit

-0.59

ãĥ

-0.59

 guid

-0.59

POSITIVE LOGITS

uish

0.73

sov

0.72

ichick

0.69

aturdays

0.67

arthed

0.66

chev

0.66

sein

0.64

 Baghd

0.62

 contemplation

0.61

ardy

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.