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

orthy

-0.72

 racks

-0.70

 Marketable

-0.70

 totem

-0.69

 Sigma

-0.67

ivals

-0.64

 Poster

-0.63

 shelves

-0.61

 Plants

-0.61

clud

-0.60

POSITIVE LOGITS

aukee

0.85

ById

0.75

href

0.71

utf

0.68

ecause

0.66

etsk

0.66

uphem

0.66

displayText

0.63

ryn

0.62

][

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.