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

 Wand

-0.66

 Graves

-0.66

 Tsukuyomi

-0.66

ollo

-0.65

 Circus

-0.64

uits

-0.63

 Cron

-0.62

cry

-0.60

 Olymp

-0.60

 Sands

-0.60

POSITIVE LOGITS

roma

0.83

ory

0.81

orical

0.78

liness

0.77

ighth

0.75

ubuntu

0.74

Reader

0.69

smanship

0.66

ĩ

0.66

resa

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.