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

ector

-0.77

cher

-0.74

 Ange

-0.67

 Mate

-0.64

 downstream

-0.64

ection

-0.63

tu

-0.62

 Chancellor

-0.61

 Raqqa

-0.61

ect

-0.60

POSITIVE LOGITS

 series

1.33

series

1.29

 Series

0.93

Series

0.88

poons

0.85

episode

0.85

isodes

0.82

 Pacers

0.78

 episodes

0.77

shows

0.75

Activations Density 0.000%

No Known Activations

This feature has no known activations.