INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

NES

-0.67

helle

-0.67

pas

-0.66

wic

-0.66

boats

-0.64

ICS

-0.63

Williams

-0.62

Bs

-0.62

eson

-0.61

oby

-0.61

POSITIVE LOGITS

icial

0.74

 ours

0.70

lore

0.68

ingen

0.68

 sorts

0.68

olon

0.67

 theirs

0.65

orsi

0.62

 Representatives

0.61

Tag

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.