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

ocene

-0.86

heit

-0.83

ropolis

-0.81

chnology

-0.80

aceous

-0.79

elin

-0.74

phasis

-0.74

romy

-0.74

iens

-0.73

alde

-0.73

POSITIVE LOGITS

é¾

0.73

 homophobic

0.70

%%%%

0.67

 Pipe

0.63

 darts

0.62

boo

0.59

 homophobia

0.59

uterte

0.58

tir

0.58

 detract

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.