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

thood

-0.78

icity

-0.66

 virginity

-0.66

adian

-0.66

dimension

-0.65

 infancy

-0.65

afer

-0.65

 livest

-0.64

 prototype

-0.63

wcsstore

-0.63

POSITIVE LOGITS

"@

0.73

Introduced

0.69

 Cosponsors

0.68

olded

0.67

âĢ¦]

0.67

Allows

0.67

cers

0.66

ONSORED

0.65

imaru

0.64

 certs

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.