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

ĸļ

-0.84

roman

-0.84

perature

-0.80

yip

-0.80

ecause

-0.75

 dilig

-0.73

orians

-0.71

atche

-0.68

oresc

-0.67

ickle

-0.66

POSITIVE LOGITS

://

0.99

 Killer

0.76

San

0.72

Su

0.72

*****

0.65

 Sense

0.64

 dens

0.64

San

0.64

holes

0.64

Dod

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.