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

-|

-0.80

GRE

-0.73

PF

-0.65

_____

-0.65

âĢ¢

-0.64

Cent

-0.64

CAP

-0.63

 Logged

-0.62

elcome

-0.62

-0.61

POSITIVE LOGITS

 asses

0.79

ulent

0.76

 stunts

0.71

 Seym

0.70

 sauces

0.70

iments

0.70

 folds

0.69

ogyn

0.69

 partName

0.67

ulence

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.