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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

SPONSORED

-0.84

arthed

-0.80

skirts

-0.73

ç«

-0.73

cedented

-0.71

á

-0.66

ä¸Ĭ

-0.65

 WATCHED

-0.63

oren

-0.63

 Continuous

-0.63

POSITIVE LOGITS

asm

0.75

olin

0.70

vati

0.70

 hypot

0.70

bc

0.67

cin

0.66

atron

0.65

iment

0.65

acan

0.65

 partly

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.