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

åĮ

-0.74

²¾

-0.69

ãĤ«

-0.69

auer

-0.67

arkin

-0.66

container

-0.65

 underestimate

-0.65

Aren

-0.63

arge

-0.62

opol

-0.62

POSITIVE LOGITS

xes

0.75

 commem

0.71

netflix

0.70

recy

0.66

merce

0.65

 seiz

0.64

urate

0.64

dule

0.64

 reconciliation

0.63

fol

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.