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

arrow

-0.77

bet

-0.71

hat

-0.70

ills

-0.69

hover

-0.69

asin

-0.66

ule

-0.66

bye

-0.66

umbered

-0.66

ensional

-0.65

POSITIVE LOGITS

REDACTED

0.69

ITY

0.68

 Lancet

0.67

 Roma

0.65

 Leth

0.65

ITIES

0.64

 millionaire

0.64

 Realms

0.63

 Shiite

0.63

 Ivory

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.