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

Ã±o

-0.72

ayne

-0.71

igible

-0.71

quickShipAvailable

-0.70

ioxide

-0.70

hyde

-0.69

WD

-0.67

lished

-0.67

CRIPTION

-0.66

eele

-0.66

POSITIVE LOGITS

iser

0.75

 mirac

0.75

uckland

0.72

 prevailed

0.67

aer

0.66

 allele

0.65

 trou

0.64

 respir

0.64

 ledge

0.63

bleacher

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.