INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

uador

-0.75

okia

-0.72

 ILCS

-0.72

regon

-0.70

umb

-0.70

trak

-0.69

soDeliveryDate

-0.69

arij

-0.68

eele

-0.67

acion

-0.67

POSITIVE LOGITS

Interstitial

0.80

ãĥĩãĤ£

0.70

feed

0.67

 Downloads

0.66

Ïĥ

0.65

usc

0.64

glers

0.64

Berry

0.64

ãĥ¼

0.63

ãĥĥãĥĪ

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.