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

Ger

-0.77

Jew

-0.72

Town

-0.66

Mist

-0.66

Val

-0.65

Film

-0.65

Austral

-0.65

Hay

-0.63

Americ

-0.63

åĤ

-0.63

POSITIVE LOGITS

axis

0.70

oÄŁan

0.68

nces

0.66

bps

0.64

onymous

0.63

agonists

0.63

chwitz

0.63

named

0.63

uru

0.62

oshenko

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.