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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Vacc

-0.70

bourg

-0.63

comed

-0.62

gp

-0.62

 Pupp

-0.62

 outbreak

-0.60

 advis

-0.60

 Rept

-0.59

 Weasley

-0.59

 ........

-0.59

POSITIVE LOGITS

ãĤ¼

0.70

alty

0.66

atory

0.66

adena

0.64

owicz

0.63

ral

0.63

IGHT

0.63

ocent

0.62

acus

0.62

izzy

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.