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

 Lose

-0.69

estyle

-0.68

ulse

-0.68

BuyableInstoreAndOnline

-0.67

inese

-0.66

KE

-0.65

anche

-0.65

 dodge

-0.65

ince

-0.65

ifle

-0.64

POSITIVE LOGITS

ĸļ

0.75

advertisement

0.75

fram

0.65

IPM

0.64

+(

0.62

hent

0.61

pixel

0.61

 branding

0.61

rett

0.61

BMC

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.