INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Sent

-0.83

RECT

-0.71

NK

-0.67

outube

-0.66

ocular

-0.64

rote

-0.64

rared

-0.63

NRS

-0.62

Appearances

-0.61

eq

-0.60

POSITIVE LOGITS

inelli

0.67

 Dusk

0.67

worth

0.63

 Junction

0.62

henko

0.62

 overlook

0.60

rowth

0.60

 Mant

0.60

thood

0.59

 Aust

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.