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

Not in Any Lists

No Comments

Negative Logits

 vape

-0.73

Ĥİ

-0.70

 lucid

-0.68

 describe

-0.68

 unsc

-0.64

 prescribe

-0.63

 shareholder

-0.63

 depictions

-0.62

 Pengu

-0.62

ãĥ¼ãĥĨ

-0.62

POSITIVE LOGITS

ebus

0.74

onial

0.73

ngth

0.71

 Conn

0.69

 Payne

0.68

itaire

0.68

breaker

0.67

 Roose

0.67

 Exile

0.66

owship

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.