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

halla

-0.71

 Coliseum

-0.70

 faculties

-0.66

gment

-0.63

ensen

-0.62

rimination

-0.61

erv

-0.61

urous

-0.60

gemony

-0.60

atana

-0.60

POSITIVE LOGITS

FAR

0.72

Attempts

0.68

OHN

0.64

OSE

0.63

Ãį

0.62

âĢł

0.62

å§«

0.62

OSED

0.61

ALS

0.61

Ware

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.