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

 Advis

-0.70

 Doodle

-0.69

 Romans

-0.68

 Bunker

-0.66

bush

-0.66

Bey

-0.66

 Mill

-0.65

screen

-0.63

 Summit

-0.63

 Fridays

-0.63

POSITIVE LOGITS

ĸļ

0.79

cedes

0.77

luaj

0.77

olen

0.72

vernment

0.71

anguages

0.71

ignty

0.70

hid

0.69

kefeller

0.68

ild

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.