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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

»Ĵ

-0.76

icago

-0.76

ħĭ

-0.72

ãĤ¼ãĤ¦ãĤ¹

-0.71

udi

-0.69

achu

-0.68

[_

-0.68

nesota

-0.67

otle

-0.67

acters

-0.64

POSITIVE LOGITS

ected

0.62

uph

0.59

 convol

0.58

bring

0.58

ru

0.57

Suz

0.57

 dynam

0.57

 Rousse

0.56

ippery

0.55

 confirmation

0.55

Activations Density 0.000%

No Known Activations

This feature has no known activations.