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

interstitial

-0.78

gets

-0.65

warts

-0.64

kson

-0.64

 sake

-0.63

atoon

-0.63

istle

-0.63

onz

-0.63

inventoryQuantity

-0.62

Â¢

-0.62

POSITIVE LOGITS

oldemort

0.67

 Bohem

0.65

 recess

0.63

ttes

0.62

gro

0.61

 Presents

0.61

released

0.61

Balt

0.61

scape

0.61

limited

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.