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

 Divinity

-0.74

 Koen

-0.70

ibrary

-0.70

 guiActiveUnfocused

-0.70

 Arch

-0.68

aceae

-0.67

 Dating

-0.66

addons

-0.66

ournal

-0.66

xual

-0.65

POSITIVE LOGITS

BIT

0.75

GH

0.72

Gs

0.72

ISA

0.72

NF

0.71

GS

0.70

KK

0.69

IDA

0.69

yr

0.69

ERG

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.