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

 Wiggins

-0.70

erness

-0.67

oggles

-0.66

nesses

-0.64

rely

-0.63

ween

-0.63

recy

-0.61

paces

-0.61

à¼

-0.60

lex

-0.58

POSITIVE LOGITS

uez

0.75

 scalp

0.74

 Tasman

0.72

isi

0.69

imar

0.66

orsi

0.65

yip

0.64

AAP

0.63

antage

0.62

Arn

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.