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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

inct

-0.73

hib

-0.72

arcer

-0.72

itus

-0.71

ocate

-0.71

ifer

-0.69

ilib

-0.69

terness

-0.68

apo

-0.67

iate

-0.66

POSITIVE LOGITS

 Gamble

0.73

backs

0.73

liner

0.70

CVE

0.63

Boe

0.62

mails

0.62

helps

0.62

bies

0.62

 cutter

0.61

Quotes

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.