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

leck

-0.85

sylv

-0.71

opes

-0.68

 sexes

-0.68

aber

-0.67

 trillions

-0.64

TPPStreamerBot

-0.63

 trillion

-0.63

iol

-0.63

billion

-0.63

POSITIVE LOGITS

 è£ıè

0.80

ronics

0.76

CPC

0.74

 constituted

0.65

 Brav

0.61

 Harbor

0.60

leneck

0.59

 Mahar

0.59

Construction

0.58

 manufacturer

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.