INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

-0.76

Justice

-0.72

âī

-0.69

ollah

-0.67

arson

-0.67

ablishment

-0.66

âķĲ

-0.66

hin

-0.65

ey

-0.64

animous

-0.63

POSITIVE LOGITS

 spont

0.76

 citiz

0.71

izoph

0.71

 taxp

0.68

 migrating

0.66

 havoc

0.66

 modelling

0.65

 mosquit

0.63

 chatting

0.62

 planners

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.