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

ģ«

-0.71

orate

-0.69

 Anyway

-0.62

 deeds

-0.61

rahim

-0.61

emonium

-0.61

 Ames

-0.60

holes

-0.60

iture

-0.59

itures

-0.59

POSITIVE LOGITS

outhern

0.72

kered

0.69

mart

0.67

inen

0.66

issan

0.65

acus

0.64

masters

0.64

tar

0.63

blade

0.63

 Maur

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.