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

 discretion

-0.65

 cogn

-0.65

URES

-0.61

 neut

-0.61

 gast

-0.61

nia

-0.60

 proportions

-0.59

Adv

-0.58

pse

-0.56

 Canaver

-0.56

POSITIVE LOGITS

obyl

0.83

tical

0.82

osher

0.74

phabet

0.71

paio

0.71

onel

0.70

 Starts

0.70

ettel

0.68

olithic

0.68

andowski

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.