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

 Eliot

-0.74

 sett

-0.69

 Niet

-0.68

 separ

-0.68

 reversible

-0.68

mar

-0.66

 bounds

-0.63

 fram

-0.62

TIM

-0.62

 orient

-0.61

POSITIVE LOGITS

rum

1.09

 partName

0.91

onda

0.78

enhagen

0.73

Cra

0.72

inventory

0.72

iosyncr

0.72

azine

0.70

ansson

0.70

elsen

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.