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

 cages

-0.68

 holes

-0.67

tips

-0.65

 proble

-0.65

 etiquette

-0.64

 seeker

-0.63

Iss

-0.60

ntil

-0.59

tical

-0.59

 saddle

-0.59

POSITIVE LOGITS

bows

0.79

 Sparkle

0.70

Joy

0.69

 Mercer

0.67

'.

0.67

emp

0.66

hig

0.64

amen

0.63

 Amin

0.63

 Marion

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.