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

vers

-0.70

 Raleigh

-0.66

 Bren

-0.66

 Army

-0.64

mits

-0.63

 Nanto

-0.63

 Nationals

-0.61

 REUTERS

-0.61

 WARN

-0.60

Fox

-0.60

POSITIVE LOGITS

geist

0.79

ortality

0.69

 Solitaire

0.68

ickle

0.68

atism

0.67

kefeller

0.66

inki

0.65

gebra

0.65

ourt

0.64

eness

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.