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

igible

-0.71

enance

-0.71

opard

-0.71

onym

-0.71

bender

-0.69

vier

-0.69

xual

-0.68

okÃ©

-0.67

 miscar

-0.65

mentation

-0.65

POSITIVE LOGITS

Cooldown

0.70

dyl

0.67

 istg

0.65

Gaming

0.63

ilon

0.63

 spaced

0.63

 Olympus

0.62

Sold

0.62

 touring

0.62

 spacing

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.