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

 Pigs

-0.87

 Silence

-0.67

 Predators

-0.65

 Sting

-0.64

 Prosecutor

-0.62

gging

-0.61

 Cameroon

-0.61

 Echoes

-0.61

SPONSORED

-0.61

 Dracula

-0.60

POSITIVE LOGITS

hesda

0.89

rentice

0.78

urat

0.71

ctic

0.70

ribune

0.66

quickShipAvailable

0.66

daq

0.66

uckland

0.66

antage

0.65

lyn

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.