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

 defect

-0.82

avorite

-0.73

ogram

-0.73

 ineligible

-0.72

hower

-0.69

 regress

-0.67

Â¶

-0.66

pse

-0.64

 snowball

-0.64

 probabilities

-0.63

POSITIVE LOGITS

 Trident

0.77

 Saban

0.75

Reviewed

0.75

wine

0.70

Ser

0.69

bard

0.67

Mek

0.66

 Punch

0.66

tainment

0.65

 Milo

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.