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

(~

-0.62

 Awakens

-0.61

idences

-0.59

estic

-0.59

 Ortiz

-0.57

NEW

-0.56

 Norton

-0.55

rikes

-0.55

 compound

-0.55

 Gibbs

-0.54

POSITIVE LOGITS

olesc

0.70

 distingu

0.66

lighting

0.66

oslov

0.66

Yel

0.65

alore

0.64

ectomy

0.63

igun

0.62

liga

0.62

 broch

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.