INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

Reviewed

-0.68

erous

-0.68

 Curve

-0.64

 Bunker

-0.64

 Continent

-0.63

tracks

-0.62

 untrue

-0.61

lists

-0.61

 Grounds

-0.61

 Sharon

-0.61

POSITIVE LOGITS

 ç¥ŀ

0.77

asin

0.74

»Ĵ

0.74

romy

0.69

edom

0.69

enez

0.66

nom

0.66

gment

0.65

çĦ

0.65

izen

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.