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

ĸļ

-0.72

wordpress

-0.67

rontal

-0.66

emp

-0.66

vc

-0.66

iffe

-0.66

 srfAttach

-0.65

emporary

-0.64

 flair

-0.64

 princ

-0.64

POSITIVE LOGITS

VPN

0.79

pread

0.74

MA

0.70

WAR

0.70

BO

0.69

DEN

0.68

Hunt

0.67

BL

0.65

kaya

0.64

ACP

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.