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

MFT

-0.78

Sources

-0.68

Aud

-0.68

Phys

-0.65

AFTA

-0.65

arde

-0.63

 Perl

-0.63

ypes

-0.62

âĢİ

-0.62

din

-0.61

POSITIVE LOGITS

ggles

0.78

aganda

0.74

ogle

0.73

choice

0.72

ceivable

0.69

usual

0.69

hift

0.68

manship

0.68

boats

0.67

eco

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.