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

elaide

-0.65

atts

-0.65

itech

-0.63

ameda

-0.63

agles

-0.63

lean

-0.63

NAS

-0.63

atti

-0.63

HAM

-0.62

ship

-0.61

POSITIVE LOGITS

 Gutenberg

0.77

Nur

0.76

rer

0.72

othe

0.70

 Kurd

0.65

luaj

0.65

oids

0.62

 Stalin

0.62

ãĥĩ

0.62

Fry

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.