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

 SHOW

-0.69

iki

-0.69

 Share

-0.68

Pri

-0.65

 Abuse

-0.62

idate

-0.60

 Fruit

-0.59

 Restaurant

-0.59

 gallery

-0.59

arya

-0.59

POSITIVE LOGITS

cffffcc

0.83

iour

0.79

graded

0.72

hered

0.71

linked

0.71

enced

0.71

mx

0.68

ordinate

0.68

scrib

0.68

atl

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.