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

 furnish

-0.68

 incl

-0.67

Ŀ

-0.63

ueller

-0.63

bind

-0.62

 estab

-0.61

inson

-0.60

adj

-0.59

á½

-0.58

 bulletin

-0.58

POSITIVE LOGITS

hiba

0.86

enegger

0.77

 mosqu

0.73

oÄŁ

0.71

 Tycoon

0.69

ync

0.68

 conflic

0.68

ilogy

0.67

uti

0.67

 Solutions

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.