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

NES

-0.68

 HOUSE

-0.67

actory

-0.66

 Folk

-0.65

NW

-0.63

 Agric

-0.63

 WHITE

-0.63

 Football

-0.63

Comb

-0.63

haw

-0.62

POSITIVE LOGITS

arat

0.69

answered

0.68

aution

0.65

unit

0.65

 circulate

0.65

esson

0.65

 weigh

0.64

 circulated

0.64

sites

0.61

Adds

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.