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

Merit

-0.79

efer

-0.74

jun

-0.74

ylum

-0.71

erv

-0.70

appro

-0.69

ction

-0.69

Af

-0.69

agall

-0.68

ney

-0.66

POSITIVE LOGITS

âĢ

0.70

ãĤ©

0.67

Hit

0.66

 outfits

0.66

 endowed

0.66

ï¸ı

0.65

 mathemat

0.64

 both

0.63

Ð´

0.62

 sqor

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.