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

opsis

-0.78

Coh

-0.70

 Bundy

-0.67

iasis

-0.64

 dominion

-0.62

 whistle

-0.62

Corp

-0.62

 Franc

-0.61

 QuÃ©

-0.60

 Quebec

-0.60

POSITIVE LOGITS

ourses

0.81

 mathemat

0.70

jriwal

0.68

vu

0.68

 traged

0.65

ustain

0.65

jri

0.64

DEV

0.62

iannopoulos

0.62

iour

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.