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

NG

-0.76

 Browne

-0.69

BG

-0.65

 beginning

-0.63

Episode

-0.61

OTT

-0.60

Rare

-0.58

Uncommon

-0.58

 Britann

-0.57

illin

-0.57

POSITIVE LOGITS

ucer

0.75

vironment

0.73

 particular

0.69

 accur

0.66

akeru

0.66

zees

0.66

alach

0.64

cipled

0.64

anooga

0.63

MpServer

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.