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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

phies

-0.72

cock

-0.72

ÂłÂłÂłÂł

-0.69

ciating

-0.67

solete

-0.66

iversary

-0.66

tainment

-0.64

Order

-0.64

Season

-0.62

tted

-0.62

POSITIVE LOGITS

 Malf

0.74

atically

0.73

backer

0.69

 seaf

0.66

 Zhou

0.66

 Ceres

0.64

 skelet

0.64

 mathematician

0.63

 Antar

0.63

 Constantin

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.