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

ulate

-0.71

brainer

-0.69

pedia

-0.67

ulated

-0.63

 narrowly

-0.62

eled

-0.62

 appliance

-0.62

pard

-0.61

 Kore

-0.61

oster

-0.59

POSITIVE LOGITS

ãĥ¼ãĥĨ

0.83

eton

0.79

ãĤ¨ãĥ«

0.79

ãĤ¤ãĥĪ

0.77

endor

0.76

 è£ıè

0.75

ãĤ±

0.72

 Clim

0.69

DragonMagazine

0.65

é¾įå¥ĳå£«

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.