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

ulhu

-0.87

 vain

-0.79

 portable

-0.65

 rubbish

-0.64

 confines

-0.60

ignt

-0.60

 signs

-0.60

 obscurity

-0.59

verning

-0.59

MAC

-0.59

POSITIVE LOGITS

bent

0.70

etheus

0.65

orp

0.65

HR

0.65

amide

0.64

 Loan

0.64

IB

0.64

andowski

0.64

ASS

0.63

 Intake

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.