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

issance

-0.72

 trick

-0.70

inis

-0.65

isd

-0.64

 substitute

-0.64

imar

-0.63

vu

-0.63

value

-0.63

mur

-0.62

?)

-0.62

POSITIVE LOGITS

ribution

0.67

Mayor

0.66

ãĥ¼ãĥ³

0.64

ements

0.64

ratulations

0.63

nesty

0.63

 mayor

0.62

pread

0.62

egg

0.61

eatures

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.