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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

iture

-0.80

 Mansion

-0.66

iates

-0.64

matically

-0.64

ingo

-0.63

 DRAGON

-0.63

 Hunting

-0.60

ucha

-0.60

 Institution

-0.60

 Unlock

-0.60

POSITIVE LOGITS

 punct

0.78

 withd

0.75

 surpr

0.74

iannopoulos

0.70

Wik

0.69

 suff

0.68

ravel

0.67

soDeliveryDate

0.64

Redditor

0.64

 contribut

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.