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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

etus

-0.73

pell

-0.68

ãĥķãĤ©

-0.67

sweet

-0.66

cess

-0.65

Gay

-0.65

XXX

-0.64

ãĥ¢

-0.64

wife

-0.63

YES

-0.63

POSITIVE LOGITS

©¶æ

0.69

 Beware

0.67

 tong

0.65

abouts

0.64

 licence

0.63

 rely

0.63

 taxpayers

0.62

CBI

0.62

 Benefit

0.61

 REUTERS

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.