INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

Joined

-0.68

 Koen

-0.67

ŃĶ

-0.64

iday

-0.63

ILLE

-0.63

 discrep

-0.63

 pigeon

-0.63

boxes

-0.63

ozyg

-0.61

 jumper

-0.61

POSITIVE LOGITS

arta

0.94

DonaldTrump

0.68

sand

0.67

abis

0.62

Reason

0.61

Merit

0.61

 Ingredients

0.60

uana

0.60

nesty

0.59

Therefore

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.