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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

foundland

-0.69

 accuser

-0.68

 elbows

-0.67

Maz

-0.64

!--

-0.62

ORN

-0.60

 smoot

-0.60

 linguistic

-0.60

 opaque

-0.60

POSITIVE LOGITS

enegger

1.01

hog

0.89

sworth

0.80

<[

0.75

icho

0.72

hiba

0.71

phies

0.70

cong

0.69

isy

0.68

itz

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.