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

ãĥ´

-0.79

ãĤ¨ãĥ«

-0.67

*/(

-0.65

ãĥ¼ãĥĨãĤ£

-0.65

 calendars

-0.64

IUM

-0.64

ãĤ¼ãĤ¦ãĤ¹

-0.64

 Protector

-0.63

 Booker

-0.63

PB

-0.63

POSITIVE LOGITS

alach

0.73

tenance

0.73

alys

0.71

alysed

0.71

unct

0.70

asca

0.68

inez

0.68

chn

0.67

abin

0.66

chwitz

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.