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

é¾įå¥ĳå£«

-0.75

bis

-0.73

 Concord

-0.70

sweet

-0.65

sac

-0.65

afort

-0.65

Russ

-0.63

Sch

-0.63

 Nanto

-0.62

çī

-0.62

POSITIVE LOGITS

icts

0.69

ãĥ£

0.62

 hemor

0.62

usercontent

0.61

gy

0.59

mA

0.59

caster

0.59

iful

0.58

resses

0.58

HIT

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.