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

©¶æ¥µ

-0.86

 mathemat

-0.81

Cyp

-0.80

 Doodle

-0.78

²¾

-0.72

 veget

-0.72

 Bleach

-0.70

ĻĤ

-0.70

 contrace

-0.69

etus

-0.69

POSITIVE LOGITS

ipeg

0.74

nesday

0.69

 Entered

0.66

bringer

0.65

Day

0.65

MY

0.65

stre

0.65

TO

0.63

vict

0.62

roud

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.