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

 McCarthy

-0.77

AUT

-0.70

 Hollande

-0.68

é¾įå

-0.66

 Monarch

-0.63

 Ãī

-0.62

 Horses

-0.62

 Sturgeon

-0.61

 Saur

-0.60

installed

-0.59

POSITIVE LOGITS

ounding

2.68

ounds

0.93

arest

0.91

ounded

0.88

querque

0.85

ensen

0.82

igning

0.77

idences

0.72

idth

0.71

onement

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.