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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Codec

-0.71

fen

-0.65

cour

-0.64

agnetic

-0.63

 Calder

-0.63

 Cele

-0.62

acan

-0.62

soever

-0.62

igham

-0.62

icho

-0.62

POSITIVE LOGITS

ernels

0.69

Attempts

0.69

Sync

0.69

 mathemat

0.64

Issue

0.63

reads

0.63

 Cosponsors

0.62

dogs

0.62

Buy

0.61

itious

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.