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

Negative Logits

markets

-0.73

GBT

-0.69

querque

-0.69

ngth

-0.67

Hau

-0.66

ãĤ¡

-0.66

nature

-0.65

obal

-0.65

endor

-0.65

Compare

-0.65

POSITIVE LOGITS

________________________________________________________________

0.65

 flavorful

0.58

 partition

0.58

Utah

0.58

than

0.56

Pie

0.56

CONT

0.56

ournal

0.56

ter

0.55

Reviewer

0.55

Activations Density 0.000%

No Known Activations

This feature has no known activations.