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

arthed

-0.76

prototype

-0.76

asher

-0.74

king

-0.73

huge

-0.72

chel

-0.70

finder

-0.70

redd

-0.69

zag

-0.69

compan

-0.69

POSITIVE LOGITS

MpServer

0.83

 Cosponsors

0.78

 Flavoring

0.70

 notwithstanding

0.70

 ante

0.70

SI

0.69

 otherwise

0.68

igham

0.68

 halftime

0.67

 msec

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.