INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

protect

-0.75

galitarian

-0.66

nings

-0.66

erver

-0.64

à¹

-0.64

writ

-0.64

 Cheong

-0.63

usting

-0.63

party

-0.63

roman

-0.63

POSITIVE LOGITS

ipment

0.64

 deliveries

0.63

 Week

0.62

flu

0.62

 Stri

0.61

 recruiting

0.60

 Devils

0.59

 Runs

0.58

 Deadline

0.58

 narrower

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.