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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ciples

-0.78

ario

-0.76

Exit

-0.75

 departure

-0.70

Ring

-0.70

catentry

-0.69

ilver

-0.66

adal

-0.66

ulo

-0.66

 DRAG

-0.66

POSITIVE LOGITS

 Represent

0.72

 Kenn

0.69

Utt

0.69

 Lenn

0.68

 Making

0.67

 Tens

0.66

Sus

0.65

orks

0.65

Ary

0.64

 Engineer

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.