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

IFrame

Link

Negative Logits

 composer

-0.72

 autom

-0.67

 recomm

-0.66

 finale

-0.65

DOC

-0.63

OND

-0.62

 Auditor

-0.61

 ambassador

-0.60

 opener

-0.59

 secretary

-0.59

POSITIVE LOGITS

xual

0.79

plet

0.74

cedented

0.74

bably

0.73

merce

0.72

crop

0.71

ppo

0.71

ngth

0.70

pta

0.70

protein

0.69

Activations Density 0.000%

No Known Activations

This feature has no known activations.