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

resident

-0.75

Container

-0.66

Compan

-0.66

Hero

-0.65

 bricks

-0.65

Record

-0.65

Heart

-0.64

Pixel

-0.62

icated

-0.62

lease

-0.61

POSITIVE LOGITS

tical

0.87

ja

0.75

ascus

0.71

agra

0.70

MpServer

0.69

imbabwe

0.69

umar

0.68

ogo

0.68

hooting

0.67

ende

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.