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

INTON

-0.66

Op

-0.65

 Salman

-0.64

 prayers

-0.64

 Medals

-0.63

PLA

-0.62

IND

-0.62

 queues

-0.61

igers

-0.60

 Winds

-0.59

POSITIVE LOGITS

kel

0.71

cial

0.69

reth

0.68

bilt

0.67

chev

0.66

loo

0.65

cester

0.63

bara

0.63

washer

0.63

sson

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.