INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

HIP

-0.71

.ï¿½

-0.70

 hello

-0.66

 laure

-0.64

_.

-0.61

âĢ

-0.60

BALL

-0.58

num

-0.58

Interstitial

-0.58

Iowa

-0.58

POSITIVE LOGITS

DragonMagazine

0.88

adian

0.78

xus

0.77

zhen

0.75

kees

0.69

osexual

0.68

raints

0.68

onds

0.67

obook

0.63

thood

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.