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

oooooooooooooooo

-0.77

 ingred

-0.74

estyles

-0.70

Firstly

-0.69

 unnecess

-0.62

awaru

-0.62

Apart

-0.61

course

-0.61

ongs

-0.61

stall

-0.59

POSITIVE LOGITS

 accordance

0.83

 Scientology

0.79

patient

0.75

 favor

0.71

versions

0.67

 Olympia

0.63

 Playboy

0.63

alter

0.63

 order

0.62

 lieu

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.